Abstract
Pancreatic cancer is an aggressive disease with nearly equal yearly rates of diagnosis and death. Current therapies have failed to improve outcomes due to rapid disease progression and late stage at presentation. Recently, pathways involved in progression and metastasis have been elucidated; however, new knowledge has not generated more effective therapies. We report on the use of subcellular fractionation and liquid chromatography (LC)-mass spectrometry to identify 3,907 proteins in four pancreatic cancer cell lines, 540 of which are unique to primary cancer cells, and 487 unique to cells derived from metastatic sites. Statistical analysis identified 134 proteins significantly differentially expressed between the two populations. The subcellular localization of these proteins was determined and expression levels for four targets were validated using western blot techniques. These identified proteins can be further investigated to determine their roles in progression and metastasis and may serve as therapeutic targets in the development of more effective treatments for pancreatic cancer.
- Pancreatic cancer
- metastasis
- subcellular proteomics
- molecular signature
- BXPC-3
- Capan-2
- SU.86.86
- Capan-1 cells
Pancreatic cancer (PC) is a highly aggressive disease, with a dismal 5-year survival rate, and is typically not diagnosed until late stages of disease progression (1). Curative therapy is limited to complete removal of the tumor at early stages; however, 80% or more of patients are not candidates for surgical resection due to advanced disease at the time of diagnosis (1, 2). In addition, PC demonstrates high levels of resistance to current modalities of chemotherapy. This resistance has been attributed, in part, to the desmoplastic nature of PC and has been postulated to inhibit blood supply and thus delivery of drugs to the site of disease (3).
Over the past decade, advances have been made in characterizing gene mutations, aberrant protein expression and cell signaling pathways that contribute to the pathogenesis of PC (3, 4). The picture that has emerged is many faceted, with multiple mutations and pathways involved in tumor progression and metastasis. In addition, the compound effects of these many factors have limited the ability of targeted approaches to increase survival rates meaningfully (3). Even when combinations of targeted therapies with standard, adjuvant and neoadjuvant therapies have been tested, little to no improvements have been demonstrated in progression-free or overall survival rates (3, 5).
Due to the fact that most PC patients have advanced disease at the time of diagnosis, these patients typically do not undergo surgery for tumor removal (6). Because of this, analysis of patient tissues from advanced and metastatic sites is rare. Nevertheless, because advanced disease is the norm for most patients, more focus needs to be placed on the study of advanced and metastatic disease in an effort to improve outcomes in such patients. In this study, we investigated differences in protein expression in PC cell lines derived from primary tumors of the pancreas compared to PC cell lines derived from metastatic sites. Utilizing subcellular fractionation combined with liquid chromatography (LC)-mass spectrometry (MS) and a statistical bioinformatics workflow, we have identified over 3,900 proteins among four cell lines. With more than 10% of identifications being unique to either the primary or metastatic cell lines, we have identified the statistically significant differential expression between the primary and metastatic tumor deposits of over 130 proteins. We believe this to be a positive step towards a better understanding of phenotypic differences in metastatic disease, which will ultimately lead to new approaches to therapy for PC patients.
Materials and Methods
Cell culture and subcellular fractionation. Pancreatic cancer cell lines BXPC-3, Capan-2, SU.86.86 and Capan-1 were obtained from the American Type Culture Collection (Manassas, VA, USA) and cultured in the recommended media under 5% CO2 at 37°C. Between passage 6 and 10, cultures with 70-80% confluence were trypsinized and centrifuged at 600 ×g for 10 min. Pellets were washed twice with 5 ml of WB washing buffer supplied in the SPEK subcellular fractionation kit (Calbiochem/Merck, Darmstadt, Germany). Subsequent steps in the SPEK subcellular fractionation followed the manufacturer’s recommended protocol for cell pellets. The resulting fractions comprised of proteins representative of the following subcellular compartments: fraction 1, cytosol; fraction 2, membrane; fraction 3, nucleus; and fraction 4, cytoskeleton. Fractions 3 and 4 for each cell line were desalted using Zeba desalting columns (Thermo Fisher Scientific, Waltham, MA, USA). Protein concentrations were determined using a 1/10 dilution of each sample in phosphate buffered saline using the Bicinchoninic Acid microplate assay (Thermo Fisher Scientific, Waltham, MA, USA).
Gel electrophoresis and in-gel tryptic digestion. 30 micrograms of each subcellular fraction for each cell line were added to 6× sample buffer containing 10% sodium dodecyl sulfate and 0.6 M dithiothreitol, boiled for 10 min and loaded onto a 10% polyacrylamide gel. Gels were run at 35 mA until the dye front was 1 cm from the bottom of the gel. Gels were then washed with de-ionized water and fixed in 50% methanol, 7% acetic acid for 15 min. Gels were then rinsed with de-ionized water, stained with Bloo-Moose Coomassie blue staining solution (KeraFAST, Winston Salem, NC, USA) for 1 h, and de-stained in two changes of de-ionized water overnight. Each gel lane was then excised from the gel, cut into 20 gel slices (0.5 cm or smaller), and each slice further de-stained and digested with trypsin, as reported previously (7).
LC-MS/MS analysis. Extracted peptides were dissolved in 20 μl resolving buffer (10% acetonitrile, 3% formic acid) and sonicated at room temperature for 5 min in a sonicating water bath, centrifuged briefly then transferred to analytical vials. Injections of 5 μl were run for each sample in duplicate using a Thermo LTQ-XL Orbitrap (Thermo Fisher Scientific) mass spectrometer, equipped with a Waters nanoACQUITY UPLC (Waters Corporation, Milford, MA, USA). Samples were separated by a 65-min linear gradient from 90% solvent I (0.1% formic acid in water)/solvent II (0.1% formic acid in acetonitrile) to 50% solvent I/II at a flow rate of 500 nl/min via reversed-phase chromatography using a trap/elute method with a C18 sample trap in line with a C18 analytical column (8). Mass spectrometric analysis included analysis of the eight most abundant ions and 30-sec dynamic exclusion.
Data analysis and compilation. Spectra were searched against the human IPI version 3.18 database using the SEQUEST search algorithm from the Bioworks software (Thermo Fisher Scientific) with the following parameters: parent mass tolerance of 10 ppm, fragment tolerance of 0.5 Da, variable modification on methionine of 16 Da, and maximum missed cleavage of 2 (9). Search results were entered into Scaffold Q+ software (Proteome Software, Portland, OR, USA) for compilation, normalization, and comparison of spectral counts, which represents the number of identified peptides (10-12).
Protein identifications were performed using minimum SEQUEST scores of: DeltaCn greater than 0.1 and Xcorr scores greater than 1.9, 2.3, 3.4 and 4.0 for singly-, doubly-, triply- and quadruply-charged peptides, respectively. Protein thresholds were set to two-peptides minimum. Single peptide hits were excluded from the analysis. The spectral count data from duplicate analyses of primary and metastatic data sets were then compared using a power law global error model (PLGEM), in order to identify statistically significant (p≤0.005) protein changes between non-tumor and tumor samples (13, 14).
Western blot analysis. Fifteen micrograms of the proteins from each subcellular fractionation were loaded onto a 10% NuPAGE gel (Invitrogen/Life Technologies, Grand Island, NY, USA) and electrophoresed at 150 V for 80 min. Proteins were transferred to a nitrocellulose membrane at 30 V for 90 min, using an Invitrogen Xcell II blot module. Transfer efficiency and evenness of sample loading were confirmed by ponceau S staining of the membrane. Membranes were blocked for 1 h at room temperature using 5% non-fat milk in tris-buffered saline containing 0.1% Tween 20 (TBST). Membranes were incubated in primary antibody at 1:1000 dilution, in blocking buffer overnight at 4°C. Blots were washed with TBST three times for 15 min and then incubated in the appropriate secondary antibody conjugated to horseradish peroxidase for 1 h at room temperature. Blots were then washed again three times each for 5 min. Chemiluminescent detection was accomplished using the ECL Prime kit from GE Healthcare Life Sciences (Pittsburgh, PA, USA) and a UVP digital imager (Upland, CA, USA). The primary antibodies used were heat shock protein 60 (HSP60), matrin-3 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anterior gradient homolog 2 (AGR2) (EMD Millipore, Philadelphia, PA, USA) and vimentin (BD Biosciences, San Jose, CA, USA).
Results
Two PC cell lines derived from primary tumors (BXPC-3 and Capan-2) and two PC cell lines derived from liver metastases (SU.86.86 and Capan-1) were investigated. The patient demographics, derivation and oncogene mutation status for each cell line are listed in Figure 1. Morphologically, three of these cell lines, BXPC-3, Capan-2 and SU.86.86 are similar in appearance, with Capan-1 exhibiting a more fibroblast-like morphology. Furthermore, each sample grouping consisted of one male and one female cell line. Both primary cell lines were derived from patients who exhibited no evidence of metastatic disease throughout their treatment (15).
Protein identifications were determined using very stringent scoring criteria, with false discovery rates of 0.2% for peptide IDs in fractions 1, 2 and 4 and 0.3% in fraction 3. Total identifications for each fraction were combined into a non-redundant list and the data are shown in Figure 2. By further combining these data into a total, non-redundant data set the final numbers showed 3,907 unique proteins which were identified with high confidence and extremely low false-discovery rates. Out of these, 2880 were common to both primary and metastatic cell lines, while 540 and 487 were unique to primary and metastatic cells, respectively. PLGEM analysis for each fraction demonstrated a number of proteins with statistically significant increases or decreases in spectral counts at the level of p≤0.005. Fraction 1 identifications included 32 differentially expressed proteins, fraction 2 had 84, fraction 3 had 16 and fraction 4 had 23. Once combined into a non-redundant list, a total of 134 proteins were differentially expressed in one or more subcellular locations between the two populations. Representative up- and down-regulated proteins are shown in Table I. Among these 134 proteins, several have been previously reported to be involved in some way in the processes of cancer progression and metastasis, whether in pancreatic cancer specifically or in other types of cancer. Specific examples are vimentin, AGR2, HSP60 and matrin-3 (16-23). We selected these four proteins for further validation using western blotting techniques. The results from the western blots (Figure 3) show a dramatic increase in vimentin expression in the SU.86.86 cells, but virtually no expression in the other cell lines, which is in close agreement with our spectral count data (data not shown). HSP60 expression on western blot demonstrated a slight increase in metastatic cells, however not dramatic enough to warrant further investigation. AGR2 demonstrated low-level expression in the nuclear fraction only of BXPC-3, but increased expression in fractions 2 and 3 in the other three cell lines, with substantially greater expression in the membrane fraction from primary, compared to metastatic cell lines. Matrin-3 was expressed at low levels in the cytoskeletal fraction of BXPC-3, demonstrated higher expression in the nuclear and cytoskeletal fractions of Capan-1 and Capan-2, and shifted to mostly nuclear expression in the SU.86.86 cell line.
Microscopic images of BXPC-3 (A), Capan-2 (B), SU.86.86 (C) and Capan-1 (D) cells in culture. Information regarding pancreatic cancer cell lines investigated: Patient demographics, derivation site and genetic profile status for commonly recognized pancreatic cancer-associated oncogenes are shown in (E). Each cell lysate was fractionated by subcellular location prior to analysis (F). KRAS: V-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog; SMAD: mothers against decapentaplegic homolog 4.
Common and unique protein identifications among primary and metastatic pancreatic cancer cells: Subcellular fractions and complete dataset.
Discussion
In this study we applied the technique of subcellular proteomics for the in-depth profiling of differential protein expression among primary and metastatic pancreatic cancer cell lines. Our investigation identified nearly 4,000 proteins, over 20% of which were unique to either primary or metastatic cell types. Further statistical analysis yielded 134 proteins demonstrating statistically significant differential expression in primary versus metastatic cell lines. As proof of principle, we selected four markers for further validation studies, two of which have been linked to PC pathogenesis in recent years.
Vimentin is a marker of epithelial-to-mesenchymal transition (EMT). The process of epithelial to EMT has been demonstrated in recent years to be closely tied to the metastatic progression of cancer. Hallmarks of EMT include a phenotypic shift in cancer cells from epithelial in nature to a more mesenchymal or fibroblast-like morphology accompanied by a decrease in E-cadherin expression and increase in vimentin expression (20, 21). Our results indicate a correlation with vimentin expression and metastasis in only one (SU.86.86) of our metastatic cell lines. This result is supported by work done by Ellenrieder et al. who also demonstrated lack of appreciable vimentin expression in Capan-1 cells (24). However, since Capan-1 cells possess mesenchymal characteristics, questions remain as to the mechanism of transition in these cells. Transient expression was of vimentin, observed by Gilles et al, in breast cancer cells during wound healing/invasion studies, with loss of vimentin expression after complete colonization of the wound site (25). However, since Capan-1 cells do not express functional mothers against decapentaplegic homolog 4 (SMAD4), a key protein in transforming growth factor beta (TGFβ)-induced EMT (26, 27), an alternative pathway to a mesenchymal phenotype is involved. Recently, Ding et al. described a role for a known cancer stem cell marker, CD133, in the EMT of Capan-1 cells (28). These two possible explanations illustrate the potentiality of additional unknown modes of metastasis. Furthermore, the questions raised underscore the need for more experiments to describe temporal and spatial protein expression during primary, metastatic and transitional stages of PC.
Western blot results comparing BXPC-3, Capan-2, SU.86.86 and Capan-1 cell lines. Subcellular fraction numbers 1, 2, 3, and 4 represent cytosolic, membrane, nuclear, and cytoskeletal fractions, respectively. HSP60: Heat shock protein 60; AGR2: anterior gradient homolog 2.
Selected differential proteins and their observed and annotated subcellular localization.
AGR2 regulates PC pathogenesis. AGR2 is an endoplasmic reticulum-resident protein which possesses protein disulfide isomerase activity. AGR2 is expressed in tissues with mucous secretion and/or endocrine function and has been shown to form mixed disulfides with intestinal mucin (16). While the full function of AGR2 within human cells is still being investigated, expression in human cancer has been widely reported to date (16). In the setting of PC, increased expression of AGR2 has been demonstrated in early neoplasms, carcinomas and PC cell lines. Additionally, secretion of AGR2 into cell culture media has been reported, and the abrogation of AGR2 expression in knockdown experiments has been shown to reduce rates of proliferation, invasion and drug resistance both in vitro and in vivo (29). Further study is ongoing with respect to the contribution of AGR2 to the metastatic progression in PC. We believe this to be a promising target for the development of new drug therapies; and that further elucidation of the pathways involved in AGR2 regulation of PC pathogenesis is warranted.
Subcellular proteomics applications. Over the past decade, clinical trials have been conducted to study the benefit of therapeutic strategies targeting molecular pathways in PC progression. Some examples include V-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS), matrix metalloproteinases (MMPs) and cyclooxygenase-2 (COX-2) inhibitors; inhibitors of the phosphatidylinositol (PI3K), V-akt murine thymoma viral oncogene homolog (AKT) and mammalian target of rapamycin (mTOR) pathways; tyrosine kinase inhibitors; and monoclonal antibodies against epidermal growth factor receptor (EGFR), human epidermal growth factor receptor 2 (HER2) and vascular endothelial growth factor (VEGF) (3). These approaches have also been combined in various strategies with standard gemcitabine regimens, gemcitabine plus radiation, and other multiple dosing configurations. Overall, the improvements in patient survival have been moderate at best, with none of these therapies providing much promise for dramatic future success (3, 5). These discouraging results emphasize the need for continued study of the disease progression of PC in an effort to identify potential candidates for more effective targeted therapy. Additionally, the challenge inherent to PC treatment that is represented by the early metastasis of this disease indicates that a better understanding of the progression to metastasis is needed. Patient-specific combinations of genetic changes also complicate the picture. A better understanding of the differences in protein expression along the continuum of progression from early stages to late metastatic disease will be helpful in designing effective, individualized therapeutic approaches based on the location within that disease continuum. In order for this to be possible, large-scale comparisons of protein expression between primary and metastatic disease states must be undertaken. Previously, we described a robust method for global proteomic profiling of PC patient tissue samples (7). Here, we have presented data demonstrating on how this method can be applied to direct proteome comparisons between primary and metastatic disease. Our data have proven that this technique is reliable for the detection of meaningful, differential protein expression between these populations. We believe that further application of this method to a larger patient cohort can be used to develop new therapeutic targets and approaches.
Acknowledgments
We thank Dr. David R. McWilliams and Kook Y. Jung in the Proteomics Laboratory at Carolinas HealthCare System for PLGEM analysis and technical assistance, respectively. This research was supported by Institutional Funds from Carolinas HealthCare System.
- Received June 7, 2012.
- Revision received July 30, 2012.
- Accepted August 1, 2012.
- Copyright © 2012 The Author(s). Published by the International Institute of Anticancer Research.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.