Proteomic analysis of differential protein expression by brain metastases of gynecological malignancies

Brain metastases of gynecological malignancies are rare, but the incidence is increasing. Patients with brain metastases have a poor prognosis, therefore early detection and optimal management is necessary. In order to determine a new biomarker, we aimed to identify proteins that associated with brain metastases. We investigated proteins associated with brain metastases of gynecological malignancies in three patients who underwent surgical resection (stage IIb cervical cancer, stage Ib endometrial cancer, and stage IIIb ovarian cancer). Proteomic analysis was performed on formalin-fixed paraffin-embedded (FFPE) samples of the primary tumors and brain metastases, which were analyzed by liquid chromatography with tandem mass spectrometry. Thereafter, candidate proteins were identified by the Scaffold system and Mascot search program, and were analyzed using western blotting and immunohistochemistry. As a result, a total of 129 proteins were identified. In endometrial and ovarian cancers, western blotting revealed that the expression of alpha-enolase (ENO1) and triosephosphate isomerase (TPI-1) was higher and the expression of Transgelin-2 (TAGLN2) was lower in metastatic tumors than in primary tumors. On the other hand, the expression of TPI-1 and TAGLN2 was lower in metastatic tumors than in primary tumors in cervical cancer. Immunohistochemistry confirmed that ENO1 expression was elevated in the metastatic tumors compared with the primary tumors. In conclusion, the present study showed that FFPE tissue-based proteomics analysis can be powerful tool, and these findings suggested that ENO1, TPI-1, and TAGLN2 may have a role in the development and progression of brain metastasis from gynecological malignancies.


Introduction
The incidence rate of gynecological malignancies is increasing in Japan, but mortality is stable due to more effective treatment and better diagnostic techniques. However, the treatment of advanced cancer remains problematic [1]. Metastasis is a major cause of morbidity and mortality in cancer patients, so investigation of the mechanisms involved is very important. Metastasis of cancer cells is a highly selective and non-random process that comprises a series of linked events. Various molecular and genetic changes define the multistep process of tumor dissemination, which has been described as the ''metastatic cascade'' [2]. Hematogenous metastasis is a complex biological process that includes the steps of intravasation, transport in the blood, extravasation, and growth in a distant organ [3]. For hematogenous metastasis to occur, every step of the cascade must be completed [4]. According to the ''seed and soil'' theory of Paget, organ selectivity of metastasis is based on the interaction between tumor cells (the ''seed'') and the microenvironment of the target organ (the ''soil''), which supports extravasation, survival, and growth of the metastatic tumor [5]. Metastasis of gynecological malignancies can occur via the lymphatic, hematogenous, and transcoelomic routes. For cervical cancer, the most frequent sites of metastasis are the lungs, paraaortic lymph nodes, supraclavicular lymph nodes, and abdominal cavity [6]. For endometrial cancer, the most frequent sites are the lung and liver, followed by other sites such as the adrenal gland, breast, bone, skin, and brain [7]. In the case of ovarian cancer, transcoelomic metastasis is the most common, followed by pelvic lymph node, peritoneal, lymphatic, and, rarely, by hematogenous spread [8]. Metastasis to the brain is one of the most feared complications of cancer, since patients with brain metastasis usually have a poor prognosis and rapidly progressive neurologic symptoms. Consequently, treatment of brain metastases is becoming an increasingly important determinant of the survival time and quality of life for cancer patients, meaning that early detection and optimal management of brain metastases are essential. According to the brain tumor registry of Japan (1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996), tumors of the lung (52.3 %), breast (8.9 %), and rectum (5.2 %) are most likely to metastasize to the brain, while brain metastasis from gynecological malignancies is rare (1.7 % for uterine cancer and 0.8 % for ovarian cancer) [9]. However, a recent study suggested that the incidence of brain metastasis from gynecological malignancies is rising along with the longer survival of patients with these tumors due to effective treatment and the availability of better imaging techniques [10].
DNA microarray analysis has now become a standard tool for molecular studies of cancer [11]. The ability to complement this approach with methods of proteomic analysis [12] is crucial for identification of proteins that may serve as targets for new antibody-based therapeutic strategies [13]. Such proteomic research is particularly important for the characterization of gene products contributing to the metastatic potential of cancer [14]. Molecular screening of metastases by proteomic analysis has been done in several previous studies, including investigations of breast cancer, hepatocellular carcinoma, and squamous cell carcinoma of the lung [15][16][17]. However, there has been little proteomic analysis of clinical samples of gynecological malignancies, including brain metastases. Therefore, the present study was performed to investigate differential protein expression in patients with brain metastases of gynecological malignancies using proteomic analysis with the hope of identifying potential new tumor markers. To do this, we performed a comparative proteomic analysis of primary and metastatic tumor tissue sample from patients with gynecological cancer by liquid chromatography with tandem mass spectrometry (LC-MS/MS).

Materials and methods
Proteomic analysis was performed on formalin-fixed paraffin-embedded samples of primary tumors and brain metastases, which were analyzed by LC-MS/MS. Candidate proteins were detected using the Scaffold system and Mascot search program. The expression of some proteins was also assessed by western blotting and immunohistochemistry.

Patients
We reviewed 15 patients with brain metastases of gynecological malignancies.

Extraction of proteins from tumor tissues
To minimize contamination of samples by stromal cells, we selected the block that contained the largest amount of tumor tissue from each patient. Extraction of crude proteins from these tissue blocks was carried out as described elsewhere [18], with minor modifications. Briefly, 10 tissue sections (each 10 lm thick) were deparaffinized in 1 mL of xylene with gentle agitation for 5 min. After removing the xylene, 1 mL of 100 % ethanol was added and the sections were agitated for 5 min. After centrifugation at 15,000g for 10 min, the supernatant was removed and the pellet was thoroughly dried under a vacuum for 10 min. Then, 100 ll of extraction buffer from the Qproteome FFPE Tissue Kit (Qiagen, Valencia, CA, USA) was added to the dewaxed tissue pellet, followed by incubation on ice for 5 min, vortex mixing, and heating at 100°C for 20 min and 80°C for 2 h in a Thermomixer at 750 rpm. After centrifugation at 14,000g and 4°C for 15 min, 10 lL of the supernatant was used to measure the protein content by the Lowry method [19].

Sample preparation and LC-MS/MS
For isolation of tissues and preparation of proteins, the Qproteome FFPE Tissue Kit (Qiagen) was employed according to the manufacturer's recommendations. After the total protein content was measured by the Lowry method, protein samples (50 lg) were divided up for LC-MS/MS analysis. These protein samples were digested with trypsin (Protease MAX Surfactant; Promega, Madison, WI, USA) and then extracted with a Zip tip C 18 pipette tip (Promega).
The resulting peptides were subjected to LC-MS/MS analysis using a capillary LC system (Magic2002; Michrom BioResources, Auburn, CA, USA) coupled to an inline nanoelectrospraymass spectrometer (LCQAdvantage; Thermo Finnegan, Waltham, MA, USA) with a silica-coated glass capillary tube (PiclTip; New Objective, Woburn, MA, USA) to obtain a peptide mass fingerprint. Raw LC-MS/MS data files were searched by both Mascot and X!Tandem for identification. To generate a statistically valid list of proteins, Scaffold was used to accommodate differences of algorithm and score calculation by the two search engines [20]. Each protein identified was assigned a biological process based on information from the international protein index (IPI) human database (European Bioinformatics Institute, 2011) and the Gene Ontology (GO) database (National Center for Biotechnology Information, 2011).

Western blotting
Protein samples (10 lg each) were mixed in sample buffer at 100°C for 5 min and subjected to SDS-polyacrylamide gel electrophoresis on 10 % polyacrylamide gels. Then, the samples were transferred to enhanced chemiluminescence membranes (ECL; Amersham-Pharmacia Biotech, Buckinghamshire, UK) that had been blocked for 1 h in ECL Advance Blocking Agent. These membranes were incubated with rabbit anti-transgelin-2 (TAGLN2) polyclonal antibody (1:200; Sigma, Saint Louis, MO, USA), rabbit anti-triosephosphate isomerase 1 (TPI-1) polyclonal antibody (1:200; GeneTex, Irvine, CA, USA), or rabbit anti-enolase 1 (ENO1) polyclonal antibody (1:250; Abcam, Cambridge, UK) for 1 h at room temperature. After washing for a total of 30 min with 5 exchanges of Tris buffered saline with Tween 20 (TBS-T), the membranes were incubated with peroxidase-labeled immunoglobulin G of the appropriate species for 1 h at room temperature. After washing a further 5 times with TBS-T, immunoreactive proteins were detected with an ECL Advance Western Blotting Detection System kit (Amersham-Pharmacia Biotech, Piscataway, NJ, USA) and an LAS-3000(Fujifilm, Tokyo, Japan). After stripping, the membranes were reprobed with mouse anti-actin monoclonal antibody (1:2,000; Sigma) as a loading control.

Immunohistochemistry
Tissue blocks of the primary tumors and brain metastases were cut into 3-lm sections, which were dewaxed, rehydrated, and incubated with 3 % hydrogen peroxide for 5 min to block endogenous peroxidases. Then, the sections were incubated with anti-ENO1 monoclonal antibody (1:100; Abcam) at room temperature for 1 h. After washing 3 times with phosphate-buffered saline, the sections were incubated with a horseradish peroxidase-labeled polymerconjugated anti-mouse secondary antibody (ENVI-SION ? ; Dako, Copenhagen, Denmark) for 30 min at room temperature. Finally, color was developed with 3,3 0diaminobenzidine tetrahydrochloride.

Proteins identified in the primary and metastatic tumors
Raw files of LC-MS/MS data were searched by both Mascot and X!Tandem for identification of proteins. To  (22) DOD (4) EC endometrial cancer, CeC uterine cervical cancer, OC ovarian cancer, CCRT concurrent chemoradiotherapy, BSO bilatelal salpingo-oophorectomy, PLN pelvic lymph node dissection, LN lymph node, WBRT whole brain radiotherapy, AWD alive with disease, DOD dead of disease a Survival from diagnosis of brain metastasis Proteomic analysis of differential protein expression 59 generate a statistically valid protein list, Scaffold was used to accommodate differences of algorithm and score calculation between the two search engines [20]. A total of 129 proteins were identified (76 in the primary tumors and 101 in the metastatic tumors) ( Table 2). Comparison of the primary and metastatic tumors revealed the differential expression of 81 proteins (28 in the solely primary tumors and 53 in the solely metastatic tumors) and the shared expression of 48 other proteins (Fig. 1a). Proteomes of the primary and metastatic tumors Each protein that we identified was assigned a biological process based on information from the IPI and GO databases to understand their role. If a protein was known to participate in more than one biological process, it was included in multiple categories. Classification according to biological processes showed that the majority of the proteins were involved in metabolic processes, developmental processes, or multicellular organismal process (Fig. 1b), with some interesting proteins that have been implicated in tumorigenesis. In metastatic tumors, proteins related to developmental and multicellular organismal process were decreased compared with primary tumors (4.7 and 3.2 %, respectively), while proteins related to metabolic processes were increased compared with primary tumors (5.3 %) (Fig. 1c).

Differential protein expression by primary and metastatic tumors
Comparison of the distribution of proteins between primary and metastatic tumors revealed differences of proteins involved in metabolic, developmental, and multicellular organismal processes (Fig. 1c). Based on the results of searches carried out in the primary and metastatic tumors, eight candidate proteins were selected (Table 3). Among these proteins, TAGLN2, TPI-1, and ENO1 were subjected to further investigation.

Confirmation of differential protein expression
To more precisely evaluate TAGLN2, TPI-1, and ENO1 expression by gynecological cancers, western blotting was carried out using proteins extracted from the 3 primary and 3 metastatic tumors. In endometrial and ovarian cancers, the expression of TPI-1 and ENO1 was higher in metastatic tumors than in primary tumors; this finding was consistent with the reported role of these proteins in promoting tumor cell survival and proliferation [21,22]. However, the expression of TAGLN2 was lower in metastatic tumors than in primary tumors; this finding was consistent with the reported role of TAGLN2 as a tumor suppressor [23]. Further, in cervical cancer, the expression of TAGLN2 and TPI-1 was lower in metastatic tumors than in primary tumors (Fig. 2). and metastatic tumors. The expression of proteins related to developmental and multicellular organismal process was lower in metastatic tumors than in primary tumors, whereas the expression of proteins related to metabolic processes was higher in metastatic tumors than in primary tumors In addition, immunohistochemical analysis was employed to assess ENO1. Figure 3 shows representative examples of immunostaining for ENO1 in primary and metastatic tumors. Both western blotting and immunohistochemistry confirmed the findings of LC-MS/MS analysis.

Discussion
The aim of this study was to identify biomarkers for brain metastases of gynecological malignancies, which have a different protein expression profile compared with primary tumors. Identification of proteins that are up-regulated or down-regulated in metastatic tumors may facilitate the detection and/or treatment of metastasis cancer. Therefore, we performed a comparative proteomic analysis of primary and metastatic gynecological cancers by LC-MS/MS analysis of proteins extracted from FFPE samples. Western blotting and immunohistochemistry were also performed to confirm the results of LC-MS/MS analysis. Until recently, protein extraction from formalin-fixed tissues was thought to be impossible because fixation by formalin creates strong intermolecular covalent bonds [24]. However, successful protein extraction protocols have been established based on the heat-induced antigen retrieval technique widely applied for immunohistochemistry and proteomic analysis by LC-MS/MS [25,26]. In the present study, the protein expression profile of metastatic tumors was compared with that of primary tumors using extracts of FFPE samples. By LC-MS/MS and scaffold analysis, 76 proteins were identified in 3 primary tumors and 101 proteins were found in 3 metastatic tumors. These proteins were related to a variety of biological processes (Fig. 1b). To find candidate proteins, the 129 proteins that we identified were divided into 15 categories based on biological processes. Comparison of the distribution of these proteins between the primary and metastatic tumors showed differences in the expression of proteins related to metabolic, developmental, and multicellular organismal processes (Fig. 1c). Eight candidate proteins were selected that were predominantly or exclusively expressed by either the primary or metastatic tumors ( Table 3). Several of the proteins identified in this study have been reported previously as possible metastasis-related proteins, including TAGLN2,  Proteins from primary and metastatic tumors were separated by SDS-PAGE and transferred to PVDF membranes, followed by detection using the respective primary antibodies and an HRPconjugated secondary antibody. In endometrial and ovarian cancers, the expression of ENO1 and TPI-1 was higher in metastatic tumors than in primary tumors, and the expression of TAGLN2 was lower in metastatic tumors than in primary tumors. Further, in cervical cancer, the expression of TPI-1 and TAGLN2 was lower in metastatic tumors than in primary tumors. P primary tumors, M metastatic tumors TPI-1, ENO1, phosphatidylethanolamine binding protein (PEBP1), and mitochondrial ATP synthase alpha-subunit (ATP5A). TAGLN2 is a poorly characterized member of the calponin family. Its closest homologue is Transgelin 1/SM22a, an actin cross-linking protein [27] that is thought to undergo down-regulation as an early marker of transformation [28]. A recent study suggested that TAGLN2 has a negative influence on metastasis suppressing the invasive capacity of tumor cells [23]. TPI-1 is an enzyme that catalyzes the reversible transformation of D-3-glyceraldehyde phosphate into dihydroxyacetone phosphate. Dihydroxyacetone phosphate is then transformed into D-3-glyceraldehyde phosphate to continue the glycolytic pathway, so TPI-1 has an important role in the process of glycolysis [29]. Changes of enzyme activity have been reported under normal and pathological conditions, and overexpression of TPI-1 may activate both energy production and protein synthesis/degradation in rapidly growing tumor cells [30].
Another study suggested that metabolic changes are associated with markedly enhanced survival and proliferation of breast cancer in the brain metastasis [21]. ENO1 is a glycolytic enzyme that catalyzes the conversion of 2-phosphoglycerate into phosphoenolpyruvate [31]. ENO1 is the frequently deregulated in various types of cancer [32]. In addition, ENO1 is more highly expressed in metastatic cancer cells compared with primary cancer cells, suggesting an oncogenic role of ENO1 [22]. PEBP1 was originally identified as an endogenous inhibitor of Raf, and it negatively regulates the Raf/MEK/ERK-signaling cascade [33]. It has been well established that PEBP1 suppresses the metastatic spread of tumor cells, and, moreover, the down-regulated expression of PEBP1 is observed in a number of human cancers [34]. ATP5A was identified in tumor metastasized to liver and was overexpressed in metastatic tumor [35]. These reports indicated that there are several changes of protein expression between primary and metastatic tumor, and the metastasis may be supported by the expression changes. In the present study of clinical tumor samples, we identified the expression of TAGLN2 in primary tumors and we also identified TPI-1 and ENO1 in metastatic tumors, predominantly by LC-MS/MS. Although the three proteins identified in this study have previously been suggested to be cancer-related, their functional role in gynecological cancer remains controversial. Also, we investigated the three proteins by western blotting. There were a few disparities in the results between western blot and LC-MS/MS. We described that TPI-1 was predominantly identified in metastatic tumors. However, the TPI-1 expression in primary tumor was higher than metastatic tumor in cervical cancer. It has been reported that TPI-1 was represented overexpression in tumor cells [30]. In addition, the overexpression of TPI-1 in metastatic tumor may not be a necessary character for metastasis. In any case, difference of cancer type may indicate difference of metastatic character. Further study is necessary to obtain their credibility as biomarkers.
Various molecular and genetic changes occur during the multistep process of tumor dissemination, and these have been called the ''metastatic cascade'' [2]. This cascade starts with escape from the primary tumor by invasion of the surrounding tissue, entry into the bloodstream (intravasation), extravasation at a distant site, and finally survival and proliferation to form metastases [36,37]. In this study, expression of TPI-1 and ENO1, which enhances cancer cell survival and proliferation, were higher in metastatic tumors than in primary tumors, and TAGLN2, which suppresses invasion of cancer cells, was lower in metastatic tumors than in primary tumors. The different expression of these proteins between primary and metastatic tumor may require characteristics for brain metastasis in these gynecological cancers. In addition, tubulin beta, 60S ribosomal protein L7 and Myosin light chain 6B were also identified in this study. Their role in tumor metastasis requires further investigation, but the role may be similar to that of proteins such as TPI-1, ENO1, and TAGLN2.
A recent study indicated that primary tumors can be regarded as genetically heterogeneous and contain subpopulations of cells with varying levels of metastatic potential [38]. The data indicate that primary tumors are subpopulations of cells with different characteristics. This study indicated, by difference of identified proteins, that cancer cells can have different characteristics between primary tumors and metastatic tumors. The different characteristics might relate to brain metastasis in this study.
In conclusion, we identified several proteins that may be involved in brain metastasis of gynecological malignancies. Since our study consisted of a small number of patients, a larger study should be conducted in order to define the biological functions and influence in tumorigenesis and metastasis. In this study, we demonstrated the biomarker analysis tool using FFPE tissue-based proteomics by comparison of expression protein. Future studies using this tool will contribute to analysis of biomarkers, not only brain metastasis but also other organ metastasis such as lung and lymph node.