Reduction of Raf-1 Kinase Inhibitor Protein Expression Correlates with Breast Cancer Metastasis

Numerous genes associated with tumorigenesis have been identified. However, although the vast majority of cancer deaths are due to metastatic disease, the identification of genes involved in metastasis has lagged behind. Of particular interest are metastasis suppressor genes, which prevent cancer cells from completing the metastatic cascade. Typically, the expression of metastasis suppressor genes is diminished in metastatic tumors compared with primary tumors. Only few metastasis suppressor genes have been positively identified (1). Recent studies have added the Raf-1 kinase inhibitor protein (RKIP) to the list of candidates for novel and clinically relevant metastasis suppressor genes. Studies in cell cultures and animal models have suggested a role of RKIP in suppressing the metastatic spread of prostate cancer, breast cancer, and melanoma cells. RKIP levels were found to be reduced or absent in metastatic variants of established cell lines derived from these cancers. Reconstitution of RKIP expression prevented invasion into the matrigel (2) and metastasis in an orthotopic prostate cancer mouse model (3) but not the growth of the primary tumors. Moreover, RKIP reexpression sensitized human prostate and breast cancer cell lines to chemotherapy-triggered apoptosis (4).

Originally, RKIP was described as a physiologic endogenous inhibitor protein of the Raf-1/mitogen-activated protein kinase (MAPK) kinase/extracellular signal-regulated kinase (ERK) pathway (5). This pathway is involved in the regulation of many fundamental cellular processes, including proliferation, differentiation, survival, and motility. At its heart is a cascade of three kinases: Raf, which phosphorylates and activates MAPK kinase and MAPK kinase, which phosphorylates and activates ERK, a serine/threonine-specific kinase with >70 known substrates (6, 7). RKIP interferes with the Raf-1-mediated phosphorylation and activation of MAPK kinase via its ability to disrupt the interaction between the two kinases (8). This pathway is altered in ∼30% of all cancers and has been implicated in invasiveness and metastasis in many in vitro models (9). The ERK pathway is also often hyperactivated in breast cancer (10). Therefore, and in the light of the growing evidence from model systems that RKIP can prevent metastasis, we examined RKIP expression and ERK activation (as measured by staining with phospho-ERK antibody) in 103 clinical samples from patients with metastatic and nonmetastatic breast cancer.

Patient samples. Human breast cancer samples were obtained from 103 patients, including 52 patients without lymph node metastases (henceforth termed node negative) and 51 patients with lymph node metastases (henceforth termed node positive). In the case of node-positive patients, the primary tumor and the corresponding lymph node metastases from the same patient (henceforth termed paired samples) were examined. Samples were obtained from patients following partial or total mastectomy, with informed consent about the use of surgically resected tumors for research purposes, according to the guidelines for research on human tissue and samples set by the University of Glasgow. Parallel samples were processed for in histologic examination by H&E staining. Prognostic indicators of tumor size, grade, histologic type, and estrogen receptor status were subsequently assessed by a breast cancer pathologist (E.M.).

Immunohistochemistry. Sections of formalin-fixed, paraffin-embedded tissue (5 μm) were deparaffinized in two changes of histoclear and rehydrated through graded alcohol to distilled water. Antigen retrieval was done using 0.01 mol/L EDTA buffer (pH 8.0) for RKIP, and 0.01 mol/L citrate buffer (pH 6.0) for phospho-ERK at 95°C for 20 minutes. RKIP protein expression was examined using a 1:1,500 dilution of polyclonal rabbit antibody raised against a recombinant full-length RKIP protein. The expression of phospho-ERK and ERK were detected using commercial rabbit antibodies (Cell Signaling Technology, Inc., Beverly, MA and Santa Cruz Biotechnology, Inc., Santa Cruz, CA, respectively), according to the manufacturer's instructions. Antibody binding was detected using the streptavidin-biotin method (Vector Avidin-Biotin Complex Elite Detection kit, Peterborough, United Kingdom) and 3,3-diaminobenzidine as chromogen (Vector 3,3-Diaminobenzidine Substrate kit). Slides were counterstained with hematoxylin, dehydrated, and mounted. Paraffin-embedded sections of prostate cancer served as positive controls for RKIP expression, and omission of the primary antibody served as negative control. The specificity of the RKIP antibody was assured by (a) replacing it with unrelated IgG and (b) preadsorbing the RKIP antibody with an excess of cognate antigen (recombinant purified RKIP produced in Escherichia coli) before use (Fig. 1).

To quantify tumor cell apoptosis, a subset of tumors from each group were stained using the In situ Cell Death Detection kit (Roche Diagnostics Ltd., Lewes, United Kingdom), according to the manufacturer's instructions. Briefly, paraffin-embedded sections were dewaxed, rehydrated to water, and incubated with 20 μg/mL proteinase K for 20 minutes at room temperature. Endogenous peroxidase activity was quenched with 1% H₂O₂, and the terminal deoxynucleotidyl transferase–mediated nick-end labeling reaction mixture was applied to sections for 1 hour at 37°C. Following several PBS washes, slides were incubated with 50 μL of converter-peroxidase for 30 minutes. DNA breakpoints were visualized using 3,3-diaminobenzidine, which stained apoptotic cell nuclei dark brown.

Evaluation of immunostaining. All slides were examined and scored by a breast pathologist (E.M.), who was blinded to both clinical and pathologic data. RKIP expression levels were scored using the visual grading system into four classes, whereby 0 = negative, 1 = weak, 2 = moderate, and 3 = intense staining. No score 3 was assigned to RKIP staining in breast cancer samples. Phospho-ERK expression was scored by assessing the average signal intensity (on a scale of 0-3) and the proportion of tumor cells staining positively (0, none; 0.1, ≤10%; 0.5, ≤50%; 1.0, >50%). The two scores were multiplied to give an overall H-score, which was considered positive if it was ≥1.0 (11). Manual scoring was confirmed using digital image analysis by the Zeiss (Thornwood, NY) Axiovision software (Randolph, NJ).

Cell culture and small interfering RNA. MCF-7 cells were maintained in DMEM (Invitrogen Ltd., Paisley, United Kingdom) supplemented with 10% fetal bovine serum at 37°C, 5% CO₂, until use. We purchased an RKIP-specific small interfering RNA (nucleotides 652-672) from Qiagen Ltd. (Crawley, United Kingdom) directed against the following RKIP target sequence (accession no. NM_002567): 5′-CCAGGCCGAGTGGGATGACTA-3′. For small interfering RNA–mediated down-regulation of RKIP, 1 × 10⁶ MCF-7 cells were transfected with a total of 2 μg plasmid DNA using the Nucleofector kit V, according to the manufacturer's instructions (AMAXA Biosystems, Cologne, Germany). Forty-eight hours after transfection, cells were starved in 0.2%FCS/DMEM for 16 hours and treated with 100 ng/mL epidermal growth factor (Roche Diagnostics). Control MCF-7 cells, also underwent serum starvation and epidermal growth factor treatment.

Protein extraction and Western blotting. Total protein extracts from MCF-7 cells were prepared by lysing cells TBS-T buffer [10 mmol/L Tris (pH 7.4), 150 mmol/L NaCl, 1% Triton 100-X], with protease inhibitors (1 mmol/L phenylmethylsulfonyl fluoride, 5 μg/mL leupeptin, 2 mmol/L sodium fluoride, 1.5 μg/mL aprotinin, 0.2 mmol/L sodium pyrophosphate, 10 mmol/L β-glycerophosphate, and 0.5 mmol/L sodium orthovanadate). Cell extracts were quantified using the bicinchoninic acid protein assay kit (Sigma-Aldrich Co. Ltd., Dorset, United Kingdom), and proteins were separated on SDS-PAGE gels and transferred onto polyvinylidene difluoride membranes. Blots were hybridized with rabbit polyclonal antibodies to RKIP (1:1,000), phospho-ERK, and ERK (all used previously for immunohistochemistry) followed by horseradish peroxidase–conjugated secondary antibodies, and bands were visualized using enhanced chemiluminescence (Amersham Biosciences UK Ltd., Buckinghamshire, United Kingdom).

Statistical analysis. The RKIP expression percentages across the tumor samples determined by the ERK activation (positive/negative) and apoptosis (positive/negative) were compared using the χ² test or Fisher's exact test. RKIP expression in the node-positive tumors, and the node-negative tumors was compared using the Mann-Whitney U test. RKIP expression in the primary tumors and the metastases paired groups was compared using the Wilcoxon signed rank test. When applicable, statistical significance is reported. Statistical analysis was done using the SPSS version 11.0 software.

Breast tissue pathology. Histologic typing of primary tumors identified ductal carcinoma as the predominant cancer type, with only occasional presence of other types, such as lobular (n = 4), mucinous (n = 1), cribriform (n = 1), and tubular (n = 3). Tumors ranged in size from 5 to 120 mm (node negative, mean = 19.95 mm) and 6 to 70 mm (node positive, mean = 23.59 mm) and represented all three stages of poorly, moderately, and well differentiated tumors.

Raf-1 kinase inhibitor protein expression in primary breast tumors. All of the primary breast tumors analyzed in this study showed RKIP immunoreactivity, except for one node-negative tumor. RKIP protein was found to be predominantly cytoplasmic, although some nuclear staining was noted. RKIP expression was observed in normal epithelial cells of the milk duct, ductal carcinoma in situ, and cancer cells. RKIP was not detectably expressed in the extracellular matrix or in the connective tissues of breast sections. The RKIP antibody was specific as an unrelated IgG control or RKIP antibody preadsorbed with recombinant RKIP protein produced in E. coli did not result in any staining (Fig. 1).

Intense RKIP staining was observed in normal milk duct epithelial cell, whereas RKIP expression was down-regulated in primary breast carcinomas (Fig. 2). The consistent yet variable down-regulation of RKIP expression in primary breast carcinomas prompted us to examine whether there was a correlation between the reduction of RKIP expression and the propensity to metastasize (Table 1A). In node-positive primary tumors, RKIP expression was moderate in 31 of 51 (60.8%) and weakly positive in 20 of 51 (39.2%) cases. In the node-negative tumors, RKIP expression was moderate in 37 of 52 cases (71.2%), weakly positive in 14 of 52 cases (26.9%), and negative in 1 of 52 cases (1.9%). This distribution of RKIP down-regulation is statistically not significant (P = 0.306) but suggestive of a trend that primary tumors with reduced levels of RKIP expression have a higher tendency to metastasize.

Raf-1 kinase inhibitor protein expression is significantly diminished in metastatic breast cancer. RKIP expression in the node-positive tumors was predominantly moderate in intensity (Fig. 3B). By contrast, in the matched lymph node metastases obtained from the same patients, RKIP expression was considerably diminished (Fig. 3E). In total, 30 of 51 cases (58.8%) were weakly positive for RKIP expression, 12 of 51 cases (23.5%) were moderately positive, and RKIP was entirely absent in a significant number of cases (9 of 51, 17.7%). This decrease of RKIP expression in metastases was found to be highly consistent and statistically significant (P = 0.000003; Table 1B), suggesting that reduction of RKIP expression is a hallmark of metastatic disease in human breast cancer.

No correlation was found between RKIP expression and established clinical and pathologic breast cancer markers, including histologic type, tumor differentiation grade, size, or estrogen receptor status (data not shown), suggesting that RKIP is independent of other markers for breast cancer progression and prognosis.

Extracellular signal-regulated kinase activation in breast cancer. As RKIP can interfere with ERK activation (5), we assessed the levels of phospho-ERK in the primary tumors and in metastases. ERK activation can be conveniently and quantitatively determined using phosphospecific antibodies that detect the activating phosphorylation sites. All tumors expressed easily detectable and rather uniformly distributed levels of total ERK (i.e., inactive and activated ERK combined; Fig. 3C). In contrast, phospho-ERK staining revealed that activated ERK was not homogeneously distributed, often only found in few tumor cells, interestingly often in cells at the invasive edges of the tumors (Fig. 3F). This staining pattern resulted in low H-scores of <1.0 for most tumors, which were thus deemed negative for phospho-ERK staining. In total, 6 of 50 (12%) node-positive tumors, 4 of 50 (8%) lymph node metastases, and 11 of 51 (21.6%) node-negative tumors reached an H-score of ≥1 and hence were scored positive for phospho-ERK (Table 2A). In immunopositive tumor cells, phospho-ERK distribution was specific and predominantly nuclear, with occasional cytoplasmic staining (Fig. 3F). Intensity ranged from weak to robust phospho-ERK immunoreactivity, and the proportion of phospho-ERK positive cancer cells varied between <5% and >80%.

No significant association between RKIP expression and phospho-ERK staining was noted for node-positive tumors or for matched lymph node metastases. Curiously, RKIP and phospho-ERK levels showed a significant association in node-negative tumors (Fisher's exact test, P = 0.027). The reason for this unexpected finding is unclear but could suggest that in breast tumors the inhibition of ERK phosphorylation may not be the relevant target for RKIP-mediated metastasis suppression (Table 2A).

Raf-1 kinase inhibitor protein enhances apoptosis in node-negative tumors. RKIP can sensitize prostate and breast carcinoma cells to drug-induced apoptosis (4). Therefore, we assessed the levels of apoptosis in a subset of 21 tumors from node-negative and node-positive groups, including paired lymph node metastases, and compared them with RKIP expression (Table 2B). No significant difference between RKIP expression and apoptosis was noted for node-positive tumors or for paired lymph node metastases. However, a statistically significant weak association between RKIP expression and apoptosis was found in node-negative tumors (Fisher's exact test, P = 0.042). This suggests that the role of RKIP to prevent metastasis may be associated with the ability to promote apoptosis.

In this study, we show that RKIP expression is significantly reduced in lymph node metastases of breast cancers. Previous studies with cultured tumor cells derived from melanoma, prostate, and breast cancers have indicated that RKIP is a strong candidate for a metastasis suppressor gene (2–4). Here, we present the first larger-scale study on clinical material from 103 breast cancer patients. Our results fully support the conclusion that RKIP is a metastasis suppressor gene, as its expression is consistently lost in lymph node metastases but not in primary tumors. Importantly, RKIP is independent of other established clinical markers, such as histologic type, tumor differentiation grade, size, or estrogen receptor status.

Our study also suggests that the requirement for a metastatic cell to reduce or lose RKIP expression is not only related to RKIP's inhibitory effect on the ERK pathway. Previous studies done in cell culture and animal models have shown that metastatic prostate cancer and melanoma cells feature elevated levels of phospho-ERK and reduced RKIP expression. Restoration of RKIP expression by exogenous expression decreased phospho-ERK levels as well as invasive behavior (2–4). However, in human breast cancer sections, ERK activation was observable in only a small number of tumors in our patient cohort and usually only in a fraction of tumor cells. ERK has been reported to be activated in breast tumors, but these results are mostly based on examining homogenized tumor tissue by Western blotting rather than by immunohistochemistry with single-cell resolution on tissue sections (10). Two recent studies have shown that ERK activation correlates with a less aggressive tumor phenotype and better survival (12, 13). Although the reasons for these discrepancies are unclear at present, in the current study, we observed that reduction in RKIP expression did not lead to activation of ERK in primary tumors or metastases. This observation could suggest that the chronic hyperactivation of ERK triggers strong feedback mechanisms that in conjunction with even low RKIP expression suffice to reduce ERK activation below the levels detectable by immunohistochemistry. ERK activation can initiate multiple feedback mechanisms, including the expression of phosphatases (MAPK phosphatases) that inactivate ERKs, and may be responsible for the poor responsiveness of ERK to mitogen activation in many cancer cell lines (14).

A more intriguing interpretation could be that RKIP, in addition to the regulation of the ERK pathway, fulfils another yet unknown function that impedes metastasis. A lead is provided by the finding that node-negative tumors exhibited a statistically significant but weak correlation between RKIP expression and apoptosis (Table 2B). It has been shown that RKIP also can inhibit the activation of the nuclear factor-κB (NF-κB) transcription factor by interfering with the degradation of its inhibitor IκB (15). NF-κB exerts a potent antiapoptotic function, is required for the development of the mammary gland, and promotes metastasis in mouse models for breast cancer (16–18). In MCF7 breast carcinoma cell lines, the down-regulation of RKIP by small interfering RNA did not affect the activation of ERK but enhanced the phosphorylation and degradation of IκB (data not shown), suggesting that the reduction of RKIP expression in mammary tumors activates the NF-κB pathway. We have tried to examine the status of NF-κB signaling in our set of breast carcinomas, but unfortunately, none of the antibodies tested (i.e., anti-IκB, anti-phospho-IκB, anti-p50, or anti-p65 NF-κB) did well in immunohistochemistry. Thus, although NF-κB certainly is a candidate, the question as to what constitutes the RKIP target relevant for metastasis needs reevaluation. We are currently addressing this question by the systematic identification of RKIP binding partners through the proteomic analysis of RKIP multiprotein complexes.

Regardless of the exact mechanism of RKIP function, the highly significant down-regulation of RKIP expression in metastases and the apparent lack of mutations in the RKIP coding sequence (2) could provide a new therapeutic target for the treatment of metastatic cancer. Preliminary observations show that RKIP protein expression can be induced by ionizing radiation⁶

We thank Oliver Rath (Signaling and Proteomics Laboratory, Beatson Institute for Cancer Research, Glasgow, UK) for kindly supplying the small interfering RNA/RKIP oligonucleotide, Margaret O'Prey and Iain Downey for technical advice, Issam Francis for assistance with the terminal deoxynucleotidyl transferase–mediated nick-end labeling assessments, and Mark Montgomery (Zeiss) for the image analysis with the Axiovision software.