Functional Clustering of Metastasis Proteins Describes Plastic

Functional Clustering of Metastasis Proteins Describes Plastic Adaptation Resources of Breast-Cancer Cells to New Microenvironments† Berta Martıń,‡,# Rebeca Sanz,‡,# Ramo ´ n Aragu ¨ e´s,§ Baldo Oliva,§ and Angels Sierra*,‡ Centre d’Oncologia Molecular, IDIBELL, Hospital Duran i Reynals, CSUB, Gran Via s/n, Km 2.7, L’Hospitalet Ll, 08907, Spain, and Grup de Bioinforma`tica Estructural (GRIB-IMIM), Universitat Pompeu Fabra, C/ Doctor Aiguader, 80 Barcelona 08003, Catalonia, Spain Received November 15, 2007

To examine the molecular mechanisms underlying breast cancer metastasis in liver and search for potential markers of metastatic progression in soft-tissue, we analyzed metastatic variants developed from the highly metastatic MDA-MB 435 cell line through in vivo stepwise selection in the athymic mice. Comparative proteomic analysis using two-dimensional electrophoresis (2DE-DIGE) revealed that 74 protein spots were reproducibly more than doubled in liver metastatic cells compared to parental counterpart. From 22 proteins identified by MALDI-TOF, belonging to intermediate filaments, intracellular transport and ATP synthesis, we generated a protein-protein interaction network containing 496 nodes, 12 of which interacted. GRP 75 was connected with four other proteins: prohibitin, HSP 27, elongin B and macropain delta chain. After functional classification, we found that pathways including hepatocyte growth factor receptor (p ) 0.014), platelet-derived growth factor (p ) 0.018), vascular endothelial growth factor (p ) 0.021) and epidermal growth factor (p ) 0.050) were predominant in liver metastatic cells, but not in lung metastatic cells. In conclusion, we suggest that GRP 75 is involved in cell proliferation, tumorigenesis and stress response in metastatic cells by recruiting signals in which the transmembrane receptor protein tyrosine kinase signaling pathway (p-value FDR ) 1.71 × 10-2) and protein amino acid phosphorylation (p-value FDR ) 3.28 × 10-2) might be the most significant biological process differentially increased in liver metastasis. Keywords: Breast cancer • Functional Pathways • Metastasis • Protein Interaction Networks • Proteomics

Introduction Despite reduced mortality in breast-cancer patients due to earlier diagnosis and implementation of adjuvant chemo- and hormone-therapies, breast cancer is still the most common cause of cancer death in women worldwide.1 Many factors and genes are involved in the initiation of breast cancer, but mortality is due to metastatic disease.2 Identification of all the genetic components that support these cell processes is a challenge for cancer cell biologist. Like other cancer entities, breast cancer often shows organ preference in metastasis formation. Bone is the most common organ in which breast carcinomas establish distant metastases (60%); it is the site about which most insight has been gained.1,3 Moreover, lung (34%) and liver (20%) are often colonized by * To whom correspondence should be addressed. Dra. Angels Sierra, Centre d’Oncologia Molecular, Institut de Recerca Oncolo`gica-IDIBELL, Hosp. Duran i Reynals, C.S.U.B., Gran Vıá, km 2,7, 08907 L′ Hospitalet de Llobregat, Barcelona, Spain. Tel, 34 93 260 74 29; fax, 34 93 260 74 26; e-mail, [email protected]. † Originally submitted and accepted as part of the “Mammary Gland and Breast Cancer Proteomics” special section, published in the April 2008 issue of J. Proteome Res. (Vol. 7, No. 4). ‡ IDIBELL. § Universitat Pompeu Fabra. # These authors contributed equally to this work

3242 Journal of Proteome Research 2008, 7, 3242–3253 Published on Web 06/27/2008

metastatic carcinoma cells, although comparable therapeutic improvement is not always achieved.4 Many of the traditional invasion pathways modulate lung metastasis. Gene expression profiling of lung metastatic sublines of a human breast-cancer cell line identified several membrane or secreted proteins that induce lung metastasis when expressed together, but not when expressed individually.5 However, little is known about whether these pathways are specific to lung versus other organs; 6 and even less is known about colonization of the liver by breast-cancer cells. Furthermore, unlike colorectal carcinomas which preferentially metastasize in liver (70%), liver involvement in breast-cancer patients is generally seen as part of more widespread metastatic disease.7 Metastasis requires a number of genetic or epigenetic alterations to complete the various steps in the migratory process. Metastasis to different organs is determined by the tumor cell phenotype and interactions between the tumor cell and the organ environment.8–10 Certain functional biological pathways from the target organ must be mimicked for cancer cells to adapt to the new microenvironment.11,12 Since genetic mutations may modify protein-signaling pathways and thus confer a survival advantage on the cell, metastasis is in the end 10.1021/pr800137w CCC: $40.75

 2008 American Chemical Society

Biological Pathways of Liver Metastasis a proteomic disease. The pathogenic signaling pathways extend to the tumor-host interface and metastatic growth depends on the proteomic tissue microenvironment. Montel and colleagues13 found that metastasis to different organs occurs through similar genetic mechanisms, indicating that the host microenvironment contributes actively to tumor progression. We hypothesize that in metastasis many phenotypes redundantly express genes that are indispensable for the metastatic process, and that metastasis diversity might be mediated by the activation of genes acting as adaptors to organspecific growth.14 Thus, the identification of nodal points on these multiple survival-signal pathways might give us useful targets for antitumor drugs and assist in metastasis diagnosis. The aims of this study were to characterize the functional phenotypes that might enhance liver metastasis in breast cancer tumors, preferentially to other foci and flowing through protein networks to analyze different and similar pathways in soft-tissue (lung-liver) metastasis. Since different pathways might be connected to the achievement of metastatic activity,15 we analyzed by two-dimensional electrophoresis (2DE-DIGE) the distinct expression of proteins between liver (435-Liver) and lung (435-Lung) metastasis variants from a well-known breast cancer cell line MDA-MB 435 (435-P). Moreover, we described the proteome network for liver and lung metastasis using a bioinformatics tool that creates and analyzes protein interaction networks (PIANA), and we classified the predominant functional phenotypes in each of these soft-tissue breast cancer metastases to characterize their phenotype through protein pathways and networks.

Materials and Methods Cells and Experimental Tissues. MDA-MB 435 cell cultures (435-P) were maintained in 1:1 (v/v) mixture of DMEM and Ham F12 medium (DMEM/F12) supplemented with 10% fetal bovine serum (FBS), 1 mM pyruvate and 2 mM L-glutamine in 5% CO2/95% air at 37 °C in a humidified incubator. We generated orthotopic primary tumors in 7-week-old athymic Nude Balb/c female mice by inoculation of 1.5 × 106 435-P cells in 0.05 mL of medium without serum in the right inguinal mammary gland (i.m.f.p.), following previous protocols.16 Tumors were excised at day 45 after cell injection, and animals were put down and organs removed, weighed and examined for metastasis at day 110. Metastatic variants in lungs were obtained as described elsewhere.17 Monolayers of lung metastatic cells were obtained from trypsin-treated histocultures (metastasis fragments of ≈1 mm3) maintained until growth in medium supplemented with 20% fetal bovine serum, 1 mg/mL penicillin and streptomycin, 1 mg/mL neomycin and 0.2 mg/ml gentamycin. To obtain liver metastasis, 1.5 × 106 435-P cells in 0.05 mL of medium without serum were inoculated intra spleen; 2 days later, spleens were removed, and 2 months later, mice were killed, the liver was removed, and we performed monolayers of liver metastatic cells, as described elsewhere.16 Sample Preparation. The proteomic analyses were performed in cultured cells derived from primary cultures of liver (435-Liver) and lung (435-Lung) metastasis and parental cells (435-P). A total of 1 × 106 cells were plated onto 75-mm flask in complete medium for 48 h, starved for 24 h in serum-free medium, and replaced by complete medium 24 h more. Cells were rinsed in ice-cold PBS prior to trypsinization with 0.05% trypsin and 0.5 mM EDTA.

research articles The cell pellets were frozen in N2 and stored at -80 °C until electrophoresis. For this purpose, 5 × 106 cells were solubilized in 200 µL of lysis buffer (LB) containing 8 M urea, 4% CHAPS (w/v), and 40 mM Tris, and then homogenized by passing through a 25-gauge needle. Insoluble material was removed by centrifugation at 13 000 rpm for 10 min at 4 °C. Protein concentration was determined by BCA Protein Assay Reagent (Pierce, Rockford, IL). Two-Dimensional Gel Electrophoresis (2DE). Protein extracts were purified using Ettan 2D CleanUp kit (Amersham Biosciences AB, Uppsala, Sweden) according to manufacturer’s instructions. A total of 50 µg of whole cell extracts of 435-P and 435-Lung or 435-Liver cells was labeled with 400 pmol of Cy3 and Cy5, respectively (Ettan DIGE, Amersham Biosciences AB) and a pool of both samples with 400 pmol of Cy2. We crossed fluorocromes between the pairs of analyzed samples to avoid differences due to staining effectiveness. Immobilized pH 4-7 gradient (IPG) 18 cm strips, (Immobiline DryStrip 4-7 NL, Amersham Biosciences AB) were rehydrated overnight in absence of sample with 350 µL of rehydration solution (8 M urea, 4% CHAPS, 0.5% IPG buffer, 13 mM DTT). Both protein samples and the pool were placed on the Immobiline Strip with the cup-loading technique. Isoelectric focusing was performed using an IPGphor apparatus (Amersham Biosciences AB) for a total of 55 kV/h, 50 µA/strip at 20 °C, according to the manufacturer’s instructions. Immediately after being focused, IPG strips were equilibrated for 15 min in 65 mM DTT, 100 mM Tris, 6 M urea, 30% glycerol, and 2% SDS, followed by 475 mM iodoacetamide, 100 mM Tris, 6 M urea, 30% glycerol, and 2% SDS. Then, the second dimension was carried out on 12.5% polyacrylamide gels of 20 × 20 cm in Protean II electrophoresis tank (Bio-Rad, Hercules, CA) for 5 h at 20 mA. For protein identification, preparative 2DE gels loaded with 300 µg of protein were run following the same protocol. Analysis of Gel Images. Three gels from three independent experiments were scanned with the Typhoon Variable Mode Imager 8600 (Amersham Biosciences AB). The digitalized 2DE gel images were standardized and compared with the DeCyder Differential Analysis Software program, version 5.0 (Amersham Biosciences AB). The ratio values were standardized, according to Ri ) log10 (V2i/V1i), where the spot volumes (Vi) were in gel sample 1 (V1i) and gel sample 2 (V2i). Ratios were expressed in the range of 2-fold increase and decrease (2 and -2). Molecular masses (MW) and isoelectric point (pI) were determined after comparison with reference gels in the SWISS2DPAGE (Expasy) database. Protein Fingerprinting by MALDI-TOF Mass Spectrometry. The selected protein spots were excised from 2DE silver-stained gels and subjected to trypsin in-gel digestion as described elsewhere18 with minor modifications. After reduction and alkylation, the gel pieces were swollen in 10 µL of a digestion buffer containing 50 mM NH4HCO3 and 5 ng/µL trypsin (modified porcine, sequencing grade, Promega, Madison, WI) in an ice bath. After 45 min, 30 µL of 50 mM NH4HCO3 was added and the mixture was left to digest overnight at 37 °C. The samples were acidified using 5% formic acid, and 10-20 µL of the supernatant containing tryptic peptides was spotted into the MALDI plate using a nanocolumn as described elsewhere19 with some modifications. A column of POROS R2 material (Applied Biosystems, Framingham, MA) was packed in a constricted GelLoader Tip (Eppendorf, Hamburg, GerJournal of Proteome Research • Vol. 7, No. 8, 2008 3243

research articles

Martıń et al.

Figure 1. Differential expression of proteins in 435-Liver and 435-Lung metastatic cells. Images of 2DE-DIGE carried out with 50 µg of total protein from 435-P and 435-Liver cells labeled with Cy5 (red) and Cy3 (green), respectively, and 435-P and 435-Lung cells labeled with Cy3 (green) and Cy5 (red), respectively. Proteins were separated on immobilized pH 4-7 gradient strips and run on 12.5% polyacrylamide gels. Gels from three different experiments were scanned using Typhoon Variable Mode Imager. The gels in the picture are representative of three independent experiments of each pared sample. They were further analyzed with the DeCyder software program version 5.0 to assess differential expression of proteins between metastasis. Proteins identified by mass spectrometry are enlarged with the boxes.

many), washed with 10 µL of acetonitrile, equilibrated with 10 µL of 5% formic acid, loaded with 10-20 µL of the sample, washed with 10 µL of 5% formic acid and eluted directly to the MALDI plate with 0.8 µL of matrix solution (20 mg/mL R-cyano4-hydroxycinnamic acid in 70% acetonitrile in 0.1% formic acid). Peptide mass fingerprinting spectra were recorded in a Voyager STR MALDI-TOF (Applied Biosystems, Framingham, MA) in positive reflector mode with delayed extraction. The spectra were analyzed with the m/z program (Genomics Solutions). Protein identification was performed against a nonredundant database (NCBI) using the MASCOT program (http://www.matrixscience.com). Experimental Validation of Specific Proteins Expressed in Metastasis. 1. Western-Blot Analysis (WB). Cells from exponential cultures were lysed in 200 µL of RIPA buffer. The separated proteins in a 12% polyacrylamide gel were transferred to PVDF membranes (Immobilon-p, Millipore Corporation, Bedford, MA). The following antibodies (Ab) were used: cathepsin D, clone C-20 (Santa Cruz Biotechnology, Santa Cruz, CA) at 1:500; Galectin 1, clone N-16 (Santa Cruz) at 1/1000; Peroxiredoxin 4, (Laboratory Frontier, Seoul, Korea) at 1/1000; GRP 75, synthetic peptide (Abcam, Cambridge Science Park, Cambrigde, U.K.) at 1/1000; HSP60, clone LK1 (Abcam) at 1/200; PCNA, clone PC10 (Santa Cruz) at 1/100; Thioredoxin-1 (Laboratory Frontier) at 1/1000; HSP 27 clone C-20 (Santa Cruz) at 1:500. Peroxidase conjugated goat anti-rabbit secondary antibody 1/2000 (Dako, DakoCytomation, Denmark), anti-mouse sec3244

Journal of Proteome Research • Vol. 7, No. 8, 2008

ondary antibody 1/2000 (Amersham Biosciences AB) or antigoat secondary antibody 1/2000 (Sigma St. Louis, MO) were used as appropriate in each case. Immunoreactive bands were viewed on BioRad GS-800 Calibrated Densitometer using the Super Signal west-Pico (Pierce). MWs were established with See Blue Plus2 prestained Stanford (Invitrogen, San Diego, CA). An anti-human actin monoclonal antibody 1/2000 (Sigma, St. Louis, MO) and anti-human tubulin R, clone B-5-1-2 (Sigma) at 1/10 000 were also used as an internal standard for densitometric analysis, which was evaluated using the quantity of a band with the Quantity One program, as the volume of the area. 2. Immunohistochemistry (IHC). Tumor and node, liver and lung metastases from mice were analyzed on paraffinembedded tissue sections of 4 µm. We incubated directly the following primary Abs: HSP 27, clone C-20 (Santa Cruz) at 1/100; cathepsin D, clone C-20 (Santa Cruz) at 1/100; Galectin 1, clone N-16 (Santa Cruz) at 1/100; HSP60, clone LK1 (Abcam) at 1/200; PCNA, clone PC10 (Santa Cruz) at 1/100; Peroxiredoxin 4 (Lab Frontier) at 1/200; GRP 75, synthetic peptide (Abcam) at 1/100. Bound antibody was viewed with appropriate biotinylated anti-IgG/anti-mouse, 1/250 (Pierce), anti-rabbit 1/250 (Vector Laboratories, Burlingame, CA) and anti-goat 1/250 (Vector Laboratories), following DAB staining and counterstaining with hematoxylin in an Olympus BX60 (Olympus Optical Co., Ltd., Japan) and digitalized in a Spot digital camera using the Spot 4.2 software (Diagnostic Instruments, Inc., Sterling Heights, MI).

research articles

Biological Pathways of Liver Metastasis Table 1. Identities of Differentially Expressed Proteins in 435-Liver and 435-Lung Metastatic Cells Swiss-Prot ID

protein name

ratio

pl

MW

location

function

Zuotin-related factor 1 Prohibitin

-9.68 9.12 65.91 Nuclear -5.23 5.57 29.86 Mitocondrial

P09382 P10599 Q99543 P00441 P28066 P56537 Q13765 Q5VU58

Galectin-1 Thioredoxin Zuotin-related factor 1 SOD1 PSMA4 p27BBP protein (eIF6) NAC alpha Tropomyosin 3

-4.69 -4.49 -3.69 -3.26 -2.89 -2.68 -2.65 -1.76

Q15370 P46108 P35232

Elongin B CRK/P38 Prohibitin

1.35 4.73 13.13 Nuclear 1.45 5.33 22.89 Cytopl/nuclear 1.68 5.57 29.86 Mitocondrial

O75832 P28072 P13693 Q13765 P38646 Q13765 P08670 Q13162 P04792 P09496-2 P15374 P07919 P09211 P00441

Gankyrin Macropain delta chain p23 NAC alpha GRP 75 NAC alpha Vimentin Prdx 4 HSP 27 Clathrin light chain A Ubiquitin thioesterase L3 Cytochrome C1. nonheme 11 kDa protein Glutathione S-Transferase P SOD1

1.69 1.88 1.97 1.98 2.13 2.40 2.64 2.65 3.36 3.59 3.95 4.32 4.69 5.69

5.71 4.80 4.84 4.52 5.87 4.52 5.06 5.86 7.83 4.43 4.84 4.39 5.09 5.70

24.43 25.36 26.85 23.34 73.92 23.34 53.52 30.75 22.43 27.08 26.18 10.74 23.43 15.80

Cytoplasmic Cytoplasmic Cytoplasmic Cytopl/nuclear Mitocondrial Cytopl/nuclear Cytoplasmic Cytopl/nuclear Cytopl/nuclear Cytoplasmic Cytoplasm Mitochondrial Mitochondrial

Glutathione S-Transferase P GRP 75 Nucleophosmin 1. isoform 2 Nucleophosmin 1 Nucleophosmin 1. isoform 2 Nucleophosmin 1 Ubiquinol-cytochrome-c reductase complex TTC1 Vimentin Nucleophosmin 1 HDGF (hepatoma derived growth factor) Histone acetyltransferase typeB Subunit2 Progesterone receptor membrane component 1 Translation elongation factor 1 beta 2 TCP1-chaperonin cofactor A HSP 60

-2.05 -1.60 -1.57 -1.55 -1.55 -1.55 -1.45 -1.42 -1.37 -1.36 -1.35 -1.35 -1.31 -1.30 -1.30 -1.23

5.44 5.97 4.47 4.64 4.47 4.64 5.49 4.78 5.06 4.64 9.19 4.89 4.56 4.50 5.25 5.70

23.45 74.02 29.62 31.09 29.62 29.62 53.27 33.53 53.71 31.09 12.82 48.13 21.77 24.92 12.90 61.35

Redox Proliferation/chaperone Transporter/coactivator Transporter/coactivator Transporter/coactivator Transporter/coactivator Electron transporter Protein folding Intermediate filament Transporter/coactivator Signal transduction DNA regulation/proliferation Progesterone receptor (?) Biosynthesis Chaperone Chaperone

P51858 P06753 O75947 P06576

HDGF (hepatoma derived growth factor) Tropomyosin 3 ATP synthase D chain mitocondrial ATP synthase H+

-1.21 -1.21 -1.85 -1.23

9.19 4.76 5.22 5.26

12.82 28.89 18.54 56.53

Mitochondrial Mitochondrial Nuclear Nuclear Nuclear Nuclear Mitochondrial Cytoplasmic Cytoplasmic Nuclear Cytoplasmic Nuclear Microsomal Cytoplasmic Chaperone Mitochondrial matrix Cytoplasmic Cytoeskeleton Mitochondrial Mitochondrial

P12004 Q03252 Q9UJZ1 P06576 P06576 P07339 Q07021 Q8NBS9

PCNA Lamin B2 STML2 ATP synthase beta subunit ATP synthase beta subunit Cathepsin D (precursor) Glycoprotein gC1qBP (p32 o p33) Thioredoxin domain containing 5 isoform 1

1.21 1.22 1.23 1.25 1.26 1.27 1.27 1.29

4.53 5.29 6.40 5.26 5.26 5.31 4.32 5.63

29.07 67.76 38.84 56.53 56.53 26.46 23.84 48.28

DNA regulation/proliferation Intermediate filament Receptor binding Synthesis ATP Synthesis ATP Protease Binding protein Redox

Q8NBS9

Thioredoxin domain containing 5 isoform 1

1.31 5.63 48.28

P07437 P13693 P07858 P27797

Beta-tubulin TCTP (p23) Cathepsin B (precursor) Calreticulin precursor (GRP60)

1.33 1.35 1.36 1.45

4.75 4.84 5.20 4.29

50.24 19.76 22.97 48.28

P28072 P54727 P08670 P27797

Multicatalytic endopeptidase complex delta chain p58 Vimentin Calreticulin precursor (GRP60)

1.45 1.51 1.53 1.54

4.80 4.79 4.99 4.29

19.59 43.17 52.46 48.28

Nuclear Nucleoplasmic Peripheral protein Mitochondrial Mitochondrial Lysosomal Mitochondrial Endoplasmic reticulum lumen Endoplasmic reticulum lumen Cytoplasmic Cytoplasmic Lysosomal Endoplasmic reticulum lumen Cytopl/nuclear Nuclear Cytoplasmic Endoplasmic reticulum lumen

LIVER Q99543 P35232

LUNG P09211 P38646 P06748-2 P06748 P06748-2 P06748 P31930 Q99614 P08670 P06748 P51858 Q16576 O00264 P24534 O75347 P10809

IHC analyses in human samples were performed on paraffinpreserved tissue sections of 4-5 µm. Antigen was retrieved by boiling samples with Na-citrate buffer (10 mM, pH 6.0) in a boiling pan 20 min. GRP 75, synthetic peptide (Abcam) at 1/100 was diluted in chemMate Antibody diluent (Dako, DakoCytomation, Denmark) for 30 min at room temperature. Detection

5.34 4.82 9.12 5.70 4.74 4.56 4.52 5.33

14.87 12.02 65.91 15.80 26.58 26.85 23.34 28.89

Cytoplasmic Cytoplasmic Nuclear Cytoplasm Cytopl/nuclear Cytopl/nuclear Cytopl/nuclear Cytoplasmic

Chaperone Regulation of proliferation/ transcription Signal transduction Redox Chaperone Redox Proteasome Translational initiator factor Biosynthesis Intermediate filament Protein binding Intracellular signaling cascade Regulation of proliferation/ transcription Proteasome Proteasome Microtubule stabilization Biosynthesis Cell proliferation Biosynthesis Intermediate filament Redox Stress Intracellular transport Hidrolase Electron transport Redox Redox

Signal transduction Intermediate filament Synthesis of ATP Synthesis of ATP

Redox Microtubes Microtubules stabilitation Protease Biosynthesis Biosynthesis DNA repair Intermediate filament Biosynthesis

used the LSAB+Kit peroxidase (Dako), visualized with DAB prior to counterstain with hematoxylin, mounted and digitalized in a Spot digital camera with the Spot 4.2 software (Diagnostic Instruments, Inc., Sterling Heights, MI). Cell Viability Assay. The MTT tetrazolium assay was used to measure cell viability, as described elsewhere.27 Briefly, 2 × Journal of Proteome Research • Vol. 7, No. 8, 2008 3245

research articles

Martıń et al.

Figure 2. Western blot analyses from metastatic variants to validate specific expression of proteins. Whole-cell lysates containing 50 µg of total protein from 435-P cells and variants that metastasize in liver (435-Lv), lung (435-L3, 435-L2/L5) and lymph nodes (435-N1, 435-N3) were loaded, separated by PAGE (A and B, acrylamide 12%; and C, acrylamide 7%) and blotted to PVDF membranes. The proteins indicated were detected by specific primary antibodies and viewed by HRP-conjugated secondary antibodies. Anti-human tubulin monoclonal antibody was used as an internal standard (semiquantitative evaluation in Table 2). Table 2. Ratios from the WB To Evaluate Each Protein Expression with Regard to the Corresponding Tubulin

435L3/435P 435L2L5/435P 435LV/435P 435N1/435P 435N3/435P

cathepsin D

galectin 1

HSP 60

GRP 75

PCNA

PRDX 4

GST-P1

HSP 27

thyoredoxin 1

1.52 2.39 1.60 0.85 1.19

0.83 0.85 0.23 0.54 0.79

0.91 1.15 0.94 1.09 1.00

0.90 1.08 1.37 1.10 1.09

1.32 1.39 1.25 1.01 0.88

1.02 1.08 0.82 1.33 1.09

0.83 0.62 0.74 0.70 0.98

0.49 0.81 0.78 0.65 0.93

0.90 0.63 0.79 0.41 0.96

103 cells/well were cultured in 96-well microtiter plates in complete medium. After 48 h, the cells were rinsed twice and incubated with serum-free media 24 h before supplementation with hepatocyte growth factor (HGF) 2.5-10 ng/mL (Sigma) for 72 h. To check cytotoxicity, cells were seeded at 7 × 103 in 96well microtiter plates in complete medium and 5-15 mM of Gleevec (Novartis, Basel) was added for 48 h. In both cases, 50 mL of MTT solution at 5 mg/mL (Sigma) was added to the cultures and the incubation continued for a further 3-h period after which 100 µL of DMSO was added. Forming formazan crystals were dissolved before measuring optical density at 540 nm on a microplate reader. GRP 75 Protein Knockdown. Stealth RNAi oligonucleotide targeting the sequence of GRP 75 as well as a nontargeting Stealth RNAi negative control (medium GC content) were obtained from Invitrogen Limited (Paisley, U.K.). The following sequence was used: 5′-CAGGACGUGAGCAGCAGAUUGUAAU3′ (NM_004134_stealth_1724). The RNA duplexes were introduced into 435-Liver cells using Lipofectamine 2000 (Invitrogen) as a transfection agent. A total of 15 × 104 cells were seeded in 6-well plates and transfected after 24 h (at 50-60% confluence). Protein knockdown was assessed by Western blot analysis 24, 48 and 72 h after transfection. The following antibodies were used: GRP 75, synthetic peptide (Abcam, Cambridge Science Park, U.K.) at 1/1000, p53 3246


clone PAb1801 (Oncogene) at 1/20, HSP 27 clone C-20 (Santa Cruz) at 1/500. Bioinformatics. Protein-protein interactions can be represented as a protein interaction network, where the nodes are proteins and the edges between the nodes are interactions between proteins. PIANA is a software framework that creates, manages and analyzes protein interaction networks. PIANA can also be used to predict new interactions for proteins, as well as identifying proteins relevant to the pathway being studied by combining data from interaction databases and electrophoresis experiments. In this study, PIANA contained 2 378 113 interactions from the database of Interacting Proteins,20 the MIPS Mammalian Protein-Protein Interaction database,21 the STRING database22 and interactions predicted by looking at structural similarities between proteins.23 We used PIANA to (1) create the protein-protein interaction network for proteins identified by mass spectrometry (hereafter referred to as root proteins); (2) predict new interactions for root proteins using the concept of interologs;24 and (3) identify proteins that connect the root proteins in the network (hereafter referred to as linker proteins). The results were analyzed by clustering the nodes of the network (proteins) according to their functions, identifying their main connectivity and their relationships with external elements (i.e., receptors) or their signaling transfer (i.e., kinases and transcription factors).

Biological Pathways of Liver Metastasis

research articles

Figure 3. Validation of different protein expression in metastasis tissues: (A) Immunohistochemistry (IHC) of paraffin-fixed metastatic tissues obtained from mice, in which 435-P breast cancer cells were injected i.m.f.p (lung and lymph node metastasis) or intraspleen (liver metastasis), and their corresponding tissues. Hematoxylin-eosin staining (H & E) of each tissue is shown: tumor (a), liver (i), lung (q), lymph node (y), viewed by light microscopy (×20). Immunolocation of galectin 1, peroxiredoxin 4, HSP 27, GRP 75, cathepsin D, HSP 60 and PCNA expression, respectively, are shown in breast tumors from mice injected with 435-P cells (b-h), in liver metastasis (j-p), lung metastasis (r-x) and lymph node metastasis from these mice (z-f′), all viewed at ×40. (B) GRP 75 expression by IHC in a representative case of paraffin-fixed liver, lung and bone metastasis from the same patient diagnosed of breast cancer. The small box in the bottom is the positive control of GRP 75 staining in colon tissue, all viewed at ×20.

Results Protein Expression Changes in Liver Metastasis and Organ Specificity. On comparing parental cells (435-P) and 435-Liver or 435-Lung cells by 2DE in three independent experiments, we detected an average of 1823 ( 25.8 spots in liver and 1780 ( 20.9 in lung analyses (Figure 1). After normalization, 931 ( 16.2 from liver and 821 ( 14.8 from lung were further analyzed to assess differential expression of proteins. The range of 2-fold increase and decrease (2 and -2) was admitted as a significant difference, with 74 liver and 71 lung protein spots differentially expressed in all three experiments. We picked out 44 liver and 51 lung protein spots that were analyzed by MALDI-TOF peptide fingerprint with a result of 22 and 38 identified proteins in liver and lung, respectively. Known protein spots are listed in Table 1, with their corresponding MW, pI and recognized function (according to SwissProt database). Proteins MW ranged from 10.738 to 74.019 kDa, and the pI from 4.29 to 9.19. The majority of liver proteins belonged to intermediate filaments, intracellular transport and ATP synthesis. Lung proteins were linked functionally to the proteasome, protein synthesis, redox and chaperones.

We chose for validation with an alternative technique proteins for which commercial antibodies were available (Supplementary Data, Table I): GRP 75, HSP 60, HSP 27, PCNA, cathepsin D, galectin 1, thioredoxin 1, peroxiredoxin 4 and glutathione S-transferase. We analyzed by WB several variants obtained by primary cultures from lung and liver metastasis (Figure 2). We validated the increased expression of GRP 75 in liver metastasis and the underexpression of galectin 1 and thyoredoxin 1 in liver metastasis, also underexpressed in lung metastasis. Furthermore, in lung metastasis, we validated the increased expression of PCNA, also increased in liver metastasis, as well as cathepsin D (Table 2.). To check protein expression changes secondary to the microenvironment, we checked by IHC their expression in experimental tissues, tumors and metastases, obtained by i.m.f.p. injection of breast carcinoma cells (Figure 3A). All the proteins analyzed were detected in tumors. The overexpression of GRP 75 and cathepsin D in liver and lung metastasis was the most relevant validation. HSP 60 was overexpressed in metastasis from liver, lung and lymph node, which suggested an increased stress response of metastatic cells at the metastatic foci that was not evident in cell extracts. Journal of Proteome Research • Vol. 7, No. 8, 2008 3247

research articles

Martıń et al.

Figure 4. A simplified protein-protein interaction network for the proteins identified by mass spectrometry (root proteins) with their linkers: (A) liver protein-protein interaction network (24), and (B) lung protein-protein interaction network (142). The small boxes show each complete network performed with 496 nodes in liver and 1338 nodes in lung.

To evaluate GRP 75 expression in human metastasis by IHC analysis, we used five sets of liver, lung and bone metastasis samples from the same patients (Figure 3B). GRP 75 was 3248


expressed preferentially in normal liver and the invasive front of four liver metastasis, showing a characteristic cytoplasmic expression with 90% of strong positive cells. Lung metastasis


research articles

Figure 5. Experimental validation of the interacting network proteins in liver metastatic cells showing expression changes of GRP 75 and HSP 27: (A) Western-blot analyses to check GRP 75 knockdown in 435-Liver metastatic cells (siGRP 75) or Stealth RNAi negative control at 72 h. Whole-cell lysates containing 50 µg of total protein were loaded, separated by PAGE in acrylamide 12% gel, and blotted to PVDF membranes. The proteins indicated were detected by specific primary antibodies and viewed by HRP-conjugated secondary antibodies. Anti-human actin monoclonal antibody was used as an internal standard. (B) Evaluation of the WB analysis with the Quantity One program using actin to protein load normalization. The histogram represents the ratio of protein expression with regard to actin.

were positive in less than 50% of cell cytoplasm in three cases. From the five cases of bone metastasis analyzed, only one had 90% of positive diffused cells. These data confirmed the high expression of GRP 75 in liver metastasis. Protein-Protein Interaction Networks in Liver and Lung Metastasis. To further analyze the functional implications of the cohort of proteins identified in each metastatic variant, protein interaction networks were elaborated for the two experimental models of breast cancer metastasis, liver and lung (hereafter named liver-network and lung-network). Differentially expressed proteins in metastatic cells with regard to the parental cells were used as root proteins to generate these networks. Root proteins identified by MALDITOF in liver (22) and in lung (30) metastatic cells generated networks containing 496 nodes (liver) and 1338 nodes (lung), respectively (Figure 4A,B, upper boxes). Twelve proteins from liver and 20 proteins from lung metastatic cells interacted though linker proteins. A simplified network containing only root and linker proteins, 24 in liver and 142 in lung, was obtained (Figure 4). In liver-network, GRP 75 was connected with four different proteins: prohibitin, HSP 27, elongin B and macropain delta chain. Moreover, GRP 75 through HSP 27 linked with RIF and vimentin was connected with p38, a member of an adapter protein family that binds to several tyrosine-phosphorylated proteins: FAK, SOS 1, PDGFR, CIN85, CBL, CAS and PI3K. In lung-network, GRP 75, nucleophosmin 1 and HSP 60 formed a nodal cluster connected with PCNA, TEF beta 2, glycoprotein gC1qBP and multicatalytic endopeptidase (Supplementary Data Table ΙΙ), including 105 different proteins altogether. To validate protein-protein interactions in liver metastasis, we chose GRP 75, which overexpression was validated with specific antibodies in cells by WB, and by IHC in experimental and in human tissues. Modulation of GRP75 expression with Stealth RNAi oligonucleotide against GRP 75 or Stealth RNAi negative control showed at 72 h a diminished protein expression which led to HSP 27 underexpression (Figure 5). However, no changes were observed on p53 expression levels. Functional Clustering of the Proteome Network Comparing Liver and Lung Metastasis. The interactome database built using the PIANA program, which allows different sources of protein-protein interactions to be integrated into a single

repository, was used to comprehend the differences between the liver and lung list of proteins. To provide a functional interpretation of the data, we classified all the proteins from the network using the FatiGO+ program, a Web-based tool for the functional profiling of genome-scale experiments that integrates heterogeneous biologically relevant information.25 FatiGO returns adjusted p-values based on three different ways of accounting for multiple testing. We used the p-value from Fisher’s exact test and the adjusted p-value calculated using the false-discovery rate (FDR), that is, the expected number of false rejections among the rejected hypothesis can be controlled.26 We used ID Ensembl identifiers as universal cross-references for 112 proteins from liver and 464 from lung to achieve the functional clustering of each metastasis. We used three different databases to classify function: BioCarta pathways (Figure 6A, Supplementary Data Table ΙΙΙ), biological process GO terms (Figure 6B, Supplementary Data Table ΙV) and molecular function GO terms (Figure 6C, Supplementary Data Table V). The biological processes predominantly involved in liver metastatic cells were transmembrane receptor protein tyrosine kinase signaling pathway (p-value FDR ) 1.71 × 10-2), protein amino acid phosphorylation (p-value FDR ) 3.28 × 10-2) and protein catabolic process (p-value FDR ) 4.76 × 10-2). This involved cell activation through protein tyrosine kinase receptors signaling pathways: hepatocyte growth factor receptor (HGFR), p ) 0.014; platelet derived growth factor (PDGF), p ) 0.018; vascular endothelial growth factor (VEGF) pathways, p ) 0.021; and epidermal growth factor (EGF), p ) 0.050. Also, the expression of genes involved in immunological processes, including cytokines and chemokines, was increased significantly in liver metastatic cells (p ) 0.001). In contrast, response to DNA damage stimulus (p-value FDR ) 3.28 × 10-2) and translation (p-value FDR ) 3.49 × 10-2) were functions associated with lung metastatic ability and were less represented in the functional clusters of proteins from the liver metastasis. Finally, significant differences of the levels of proteins implicated in the inhibition of cellular proliferation by Gleevec was found in liver versu. lung metastatic cells (p ) 0.050). To validate de different functions described in liver or lung metastatic cells, we treated them with HGF, ligand of HGFR, and with Gleevec, inhibitor of the tyrosine kinase activity, to check its cytotoxic effect (Figure 7). Journal of Proteome Research • Vol. 7, No. 8, 2008 3249

research articles

Martıń et al.

Figure 6. Histograms show the comparative analysis between liver and lung functional clusters from different database: (A) FatiGO: Biocarta pathways; (B) FatiGO: Biological process GO terms; (C) FatiGO: Molecular function GO terms. The number of proteins included in each functional group and the significance of the statistical difference is shown. 3250



research articles preventing the rapid rise in mitochondrial reactive oxygen species.30 GRP 75 and HSP 60 are like protein brothers with distinctive structural characteristics and behavior, with lifeessential roles in the mitochondria and which function in a different nonredundant manner.31 One subunit of HSP 60 and GRP 75 was underexpressed in lung metastatic cells 2DE experiments; however, we found the overexpression of both in lung and liver experimental metastasis, suggesting that these chaperones might cooperate in various mitochondrial processes. Furthermore, we reported the overexpression of HSP 60 in primary breast carcinomas that metastasized in lung and liver. This indicated a role for both proteins as a general metastatic competency genes acting as nodal points on the multiple survival-signal pathways needed to achieve metastatic growth.32 In agreement with these results, it has been reported overexpression of HSP 60 and cathepsin D in a different metastatic variant from MDA-MB 435 cells.33 In corroboration of these findings, GRP 75 overexpression has been described in hepatocellular carcinoma as closely associated with advanced tumor stages and metastatic progression. Its implications for increased malignancy and aggressive tumor behavior suggest it could be a tumor marker for predicting early recurrence.34

Figure 7. Validation of different functions described in silico in liver or lung metastatic cells: (A) viability of 435-P, 435-Liver and 435-Lung cells treated 72 h with HGF (2.5-10 ng/mL). The growth of liver metastatic cells increased in the presence of HGF (2.5-10 ng/mL) with regard to lung metastatic cells, p ) 0.040 (*). These results are representative of results obtained in one of three experiments, and are expressed as a percentage of absorbance in cells treated with regard to nontreated cells. (B) Viability of 435-P, 435-Liver and 435-Lung cells treated 48 h with Gleevec (5-15 µM). Cytotoxicity is higher in lung than in liver metastatic cells, p ) 0.050 (*), p ) 0.001 (**) at 10 µM and 15 µM, respectively. These results are representative of results obtained in one of three experiments, and are expressed as a percentage of absorbance in cells treated with regard to nontreated cells.

The growth of liver metastatic cells increased in the presence of HGF with regard to parental or lung metastatic cells (p ) 0.040). Moreover, Gleevec cytotoxicity is higher in liver than in lung metastatic cells (p ) 0.001). Thus, the different proteome network define different signaling activation in liver and lung metastatic cells, with the result of Gleevec resistance in the lung metastatic variant.

Discussion This study showed some important differences in biological processes involved in liver and lung metastatic variants from the same breast cancer cells, which have different functional phenotypes at each metastatic location, with consequences for seeding and growth at a secondary site and for therapy.27 The most reproducible characteristic of liver metastatic cells was the overexpression of GRP 75, increased in 2DE experiments, in ex vivo liver metastasis obtained from the experimental model in nude mice, and in liver metastasis from patients. GRP 75 or mortalin is a member of the heatshock protein 70 family involved in the mitochondrial chaperone machine. By interacting with p53, GRP 75 regulates the machine’s function; as such, GRP 75 is involved in pathways that regulate cell proliferation, tumorigenesis and stress response.28,29 GRP 75 expression extends the in vivo life span, suppressing apoptosis from various stressors by

Transmembrane receptor protein tyrosine kinase signaling and protein aminoacid phosphorylation pathways were the biological processes more significantly involved in liver metastasis compared with lung. In particular, p38 showed differential expression in liver metastasis. P38 can recruit signals from PDGFR, VGFR, EGFR and HGFR modulating the action of GRP 75, which in turns induces immortality and/ or tumorigenicity by inactivation of p53.30,35 GRP 75 and raf-1 compete for binding to BAG-1 in fibroblast, suggesting that growth- and/or survival-promoting effect of BAG-1 may be mediated by activation of raf-1-dependent mitogenactivated protein kinase pathways and negatively regulated by chaperone binding.36 GRP 75 interacts with prohibitin restricted to the mitochondria where it might have a role in the maintenance of mitochondrial function,37 plus transcriptional functions in the nucleus, enhancing p-53 transcriptional activation. We also found significant differences in proteins related to Gleevec treatment response (p ) 0.050) in liver, compared with lung metastatic cells. This drug traditionally used in chronic myelogenous leukemia acts by inhibiting oncogenic BCR-ABL tyrosine kinase and is also a potent tyrosine kinase inhibitor of PDGF-R.38,39 These data suggest that activated tyrosine kinase receptors in breast cancer cells are important in the development of soft-tissue metastasis in breast cancer patients and a novel target for therapy to prevent liver metastasis. DNA damage stimulus and translation were the biological processes involved in lung metastasis that were different in liver metastasis. Consequently, molecular functions related to the structural constituents of the ribosome were much greater in lung metastatic cells than in liver ones. These data reinforce the idea that microenvironment can induce an adaptation process in metastatic cells, by selecting the most important pathways to survive in secondary organ conditions, probably in mimicry of the invaded organ’s normal major functions. Moreover, the similarity between lung and liver proteome could indicate that liver metastasis in breast cancer patients is generally seen as part of a more widespread burden of metastatic disease.7 Journal of Proteome Research • Vol. 7, No. 8, 2008 3251

research articles The MDA-MB-435 human breast cancer cell line and their metastatic variants have served as the mainstay of soft-tissue metastasis analysis. The tissue of origin has been a matter of debate since analysis of DNA microarray data indicated that MDA-MB-435 might be of melanocytic origin, due to their similarity to melanoma cell lines.40 Since our results provide information about the pathogenesis of soft tissue metastasis, the conclusions can be extrapolated regardless of the primary tumor location. Organ-specific sites of metastatic lesions are determined, in part, by adhesive interactions of malignant cells with organ microvessel endothelial cells (EC) and underlying extracellular matrix (ECM). When different organs are compared, the expression in liver metastases differs most from other metastatic sites and primary tumors, possibly due to organ-specific angiogenic and lymphangiogenic responses to metastasisrelated hypoxia.41 In fact, the distinctly different but complementary metabolic function of tumor cells, tumor-associated fibroblast, and vessels shows a complex spatial organization compatible with synergistic support to each other.42 The capacity for understanding biological complexity is often limited by the ability to define relevant phenotypes. Although the ability to interpret the meaning of the individual proteins in these signatures remains a challenge, this does not diminish the power of the signature to characterize biological states. Further analysis designed to show in a cohort of primary breast tumors the value of GRP 75 protein in liver metastasis diagnosis may result in patient prognisis being improved by preventive therapies. However, the response of a cell to the myriad of signals that it receives is varied, and depends on many factors. The most frequently studied responses involve growth factor signaling and these signaling cascades have become key targets for cancer therapy. Moreover, experimental in vivo analysis to check the cause/effect of GRP 75 in soft-tissue metastasis could result in a new therapeutic targets. Abbreviations: FDR, false discovery rate; GRP, glucoseregulated protein; HSP, heat shock protein; IHC, immunohistochemistry; PIANA, bioinformatic program for protein interaction network analyses; i.m.f.p., right inguinal mammary gland; 2DE, two-dimensional electrophoresis; WB, Western blot analysis.

Acknowledgment. We thank Ma Carmen Arriba and Naiara Santana for their collaboration. We are grateful to Mr. R. Rycroft for expert language advice. We acknowledge all the partners of the METABRE consortium for their collaboration and stimulating criticism: M. Bracke (Ghent University Hospital, Belgium), R. Buccione (Consorzio mario Negri Sud, Italy), Bellahce`ne and V. Castronovo (University of Lie`ge, Belgium) for providing breast cancer and metastasis samples, P. Cle´ment-Lacroix (Prostrakan, France), P. Cle´zardin (INSERM, France), S. Eccles (Institute of Cancer Research, U.K.), R. Lidereau (Centre Rene´ Huguenin, France), A. Teti (University of l’Aquila, Italy), M. Ugorski (Wroclaw Agriculture University, Poland), and G. van der Pluijm (Leiden University Medical Center, The Netherlands). The authors thank Ms. Cristina Chiva at the Proteomics Unit, the Pompeu Fabra University, for her expert technical assistance. This study was supported by grants from the Ministry of Health and Consumer Affairs FIS/PI041937 and FIS/PI071245, by the EC MetaBre contract no. LSHC-CT-2004-506049, and by the Ministry of Education and Science SAF2004-0188-E. B.O. and 3252


Martıń et al. R.A. acknowledge grants from Ministry of Education and Science (MEC BIO2005-00533 and MCyT BIO2002-0369).

Supporting Information Available: Table Ι, antibodies used to validated protein expression in cell extracts (WB) and tissues (IHC); Table 1I, root and linker proteins extracted from the network obtained with the root protein, proteins underlined are associated with at least one of these keywords: ‘tumor’, ‘onco*’, ‘cancer’, ‘apoptosis’, ‘death’, ‘proliferation’; Table III, list of proteins from classified functions with BioCarta pathways (Figure 5A); Table IV, list of proteins from classified functions with biological process GO terms (Figure 5B); Table V, list of proteins from classified functions with molecular function GO terms (Figure 5C); Figure 1, Western-blot analyses of 2DE from MDA-MB 435 cells. Fifty micrograms of protein extracts was applied to immobilized pH 4-7 gradient, and isoelectric focusing was performed according to manufacturer’s instructions. The second dimension was carried out on 12.5% polyacrylamide gels and blotted to PVDF membranes. Glutathione S-transferase P (GSTP-PI), cathepsin D, HSP 27 and vimentin were detected using specific primary antibodies and viewed with HRPconjugated secondary antibodies. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) National Institutes of Health Consensus Development Panel. J. Natl. Cancer Inst. Monogr. 2001, 30, 5-15. (2) Sorlie, T. Eur. J. Cancer 2004, 40 (18), 2667–2675. (3) Hortobagyi, G. N. Cancer 2000, 88 (12 Suppl.), 3073–3079. (4) Kaal, E. C.; Niel, C. G.; Vecht, C. J. Lancet Neurol. 2005, 4 (5), 289– 98. (5) Minn, A. J.; Gupta, G. P.; Siegel, P. M.; Bos, P. D.; Shu, W.; Giri, D. D.; Viale, A.; Olshen, A. B.; Gerald, W. L.; Massague, J. Nature 2005, 436 (7050), 518–524. (6) Steeg, P. S. Nat. Med. 2006, 12 (8), 895–904. (7) Schluter, K.; Gassmann, P.; Enns, A.; Korb, T.; Hemping-Bovenkerk, A.; Holzen, J.; Haier, J. Am. J. Pathol. 2006, 169 (3), 1064–1073. (8) Liotta, L. A.; Kohn, E. C. Nature 2001, 411 (6835), 375–379. (9) Fidler, I. J. Nat. Rev. Cancer 2003, 3 (6), 453–458. (10) Lee, B. C.; Lee, T. H.; Avraham, S.; Avraham, H. K. Mol. Cancer Res. 2004, 2 (6), 327–338. (11) Chambers, A. F.; Groom, A. C.; MacDonald, I. C. Nat. Rev. Cancer 2002, 2 (8), 563–572. (12) Pantel, K.; Brakenhoff, R. H. Nat. Rev. Cancer 2004, 4 (6), 448– 456. (13) Montel, V.; Huang, T. Y.; Mose, E.; Pestonjamasp, K.; Tarin, D. Am. J. Pathol. 2005, 166 (5), 1565–1579. (14) Espana, L.; Martin, B.; Aragues, R.; Chiva, C.; Oliva, B.; Andreu, D.; Sierra, A. Am. J. Pathol. 2005, 167 (4), 1125–1137. (15) Cheng, J. D.; Weiner, L. M. Clin. Cancer Res. 2003, 9 (5), 1590– 1595. (16) Fernandez, Y.; Espana, L.; Manas, S.; Fabra, A.; Sierra, A. Cell Death Differ. 2000, 7 (4), 350–359. (17) Mendez, O.; Fernandez, Y.; Peinado, M. A.; Moreno, V.; Sierra, A. Clin. Exp. Metastasis 2005, 22 (4), 297–307. (18) Shevchenko, A.; Wilm, M.; Vorm, O.; Mann, M. Anal. Chem. 1996, 68 (5), 850–858. (19) Gobom, J.; Nordhoff, E.; Mirgorodskaya, E.; Ekman, R.; Roepstorff, P. J. Mass Spectrom. 1999, 34 (2), 105–116. (20) Salwinski, L.; Miller, C.; Smith, A.; Pettit, F.; Bowie, J.; Eisenberg, D. Nucleic Acids Res. 2004, 32, D449–451. (21) Mewes, H. W.; Amid, C.; Arnold, R.; Frishman, D.; Guldener, U.; Mannhaupt, G.; Munsterkotter, M.; Pagel, P.; Strack, N.; Stumpflen, V.; Warfsmann, J.; Ruepp, A. Nucleic Acids Res. 2004, 32 (Database issue), D41-44. (22) von Mering, C.; Huynen, M.; Jaeggi, D.; Schmidt, S.; Bork, P.; Snel, B. Nucleic Acids Res. 2003, 31 (1), 258–261. (23) Espadaler, J.; Romero-Isart, O.; Jackson, R. M.; Oliva, B. Bioinformatics 2005, 21 (16), 3360–3368. (24) Yu, H.; Luscombe, N. M.; Lu, H. X.; Zhu, X.; Xia, Y.; Han, J. D.; Bertin, N.; Chung, S.; Vidal, M.; Gerstein, M. Genome Res. 2004, 14 (6), 1107–1118.

research articles

Biological Pathways of Liver Metastasis (25) Al-Shahrour, F.; Minguez, P.; Tarraga, J.; Medina, I.; Alloza, E.; Montaner, D.; Dopazo, J. Nucleic Acids Res. 2007, 35 (Web Server issue), W91–96. (26) Al-Shahrour, F.; Dıáz-Uriarte, R.; Dopazo, J. Bioinformatics 2004, 20 (4), 578–580. (27) Gu, B.; Espana, L.; Mendez, O.; Torregrosa, A.; Sierra, A. Carcinogenesis 2004, 25 (12), 2293–2301. (28) Wadhwa, R.; Yaguchi, T.; Hasan, M. K.; Taira, K.; Kaul, S. C. Biochem. Biophys. Res. Commun. 2003, 302 (4), 735–742. (29) Kaul, S. C.; Aida, S.; Yaguchi, T.; Kaur, K.; Wadhwa, R. J. Biol. Chem. 2005, 280 (47), 39373–39379. (30) Liu, Y.; Liu, W.; Song, X. D.; Zuo, J. Mol. Cell. Biochem. 2005, 268 (1-2), 45–51. (31) Deocaris, C. C.; Kaul, S. C.; Wadhwa, R. Cell Stress Chaperones 2006, 11 (2), 116–128. (32) Sanz, R.; Aragu ë´s, R.; Stresing, V.; Martıń, B.; Landemaine, T.; Oliva, B.; Keltouma; Driouch, K.; Lidereau, R.; Sierra, A. Clin. Exp. Metastasis 2007, 24, 673–683. (33) Li, D. Q.; Wang, L.; Fei, F.; Hou, Y. F.; Luo, J. M.; Zeng, R.; Wu, J.; Lu, J. S.; Di, G. H.; Ou, Z. L.; Xia, Q. C.; Shen, Z. Z.; Shao, Z. M. Proteomics 2006, 6 (11), 3352–3368. (34) Yi, X.; Luk, J. M.; Lee, N. P.; Peng, J.; Leng, X.; Guan, X. Y.; Lau, G. K.; Beretta, L.; Fan, S. T. Mol. Cell. Proteomics 2007, 7 (2), 315– 325.

(35) Wadhwa, R.; Takano, S.; Robert, M.; Yoshida, A.; Nomura, H.; Reddel, R. R.; Mitsui, Y.; Kaul, S. C. J. Biol. Chem. 1998, 273 (45), 29586–29591. (36) Townsend, P. A.; Cutress, R. I.; Sharp, A.; Brimmell, M.; Packham, G. Cancer Res. 2003, 63 (14), 4150–4157. (37) Misishra, S.; Murphy, L.; Nyomba, G.; Murphy, L. Trends Mol. Med. 2005, 11 (4), 192–197. (38) Wang, R. F.; Yokoi, K.; Bucana, C. D.; Tsan, R.; Killion, J. J.; Evans, D. B.; Fidler, I. J. Clin. Cancer Res. 2003, 9 (17), 6534–6544. (39) Kitadai, Y.; Sasaki, T.; Kuwai, T.; Nakamura, T.; Bucana, C. D.; Fidler, I. J. Am. J. Pathol. 2006, 169 (6), 2054–2065. (40) Ross, D. T.; Scherf, U.; Eisen, M. B.; Perou, C. M.; Rees, C.; Spellman, P.; Iyer, V.; Jeffrey, S. S.; Van de Rijn, M.; Waltham, M.; Pergamenschikov, A.; Lee, J. C.; Lashkari, D.; Shalon, D.; Myers, T. G.; Weinstein, J. N.; Botstein, D.; Brown, P. O. Nat. Genet. 2000, 24 (3), 227–235. (41) Weigelt, B.; Peterse, J. L.; van ’t Veer, L. J. Nat. Rev. Cancer 2005, 5 (8), 591–602. (42) Koukourakis, M. I.; Giatromanolaki, A.; Harris, A. L.; Sivridis, E. Cancer Res. 2006, 66 (2), 632–637.

PR800137W

Journal of Proteome Research • Vol. 7, No. 8, 2008 3253

Functional Clustering of Metastasis Proteins Describes Plastic

Recommend Documents