Proteomics Analysis of H-RAS-Mediated Oncogenic Transformation in a Genetically Defined Human Ovarian Cancer Model

Proteomics analysis of H-RAS-mediated oncogenic transformation in a genetically deﬁned human ovarian cancer model

Travis Young1, Fang Mei1, Jinsong Liu2, Robert C Bast Jr3, Alexander Kurosky4 and Xiaodong Cheng*,1

1Department of Pharmacology and Toxicology, The University of Texas Medical Branch, 301 University Boulevard, Galveston, TX 77555-1031, USA; 2Department of Pathology, The University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030, USA; 3Department of Experimental Therapeutics, The University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030, USA; 4Department of Human Biological Chemistry and Genetics, The University of Texas Medical Branch, 301 University Boulevard, Galveston, TX 77555-1031, USA

RAS is a small GTP binding protein mutated in Introduction approximately 30% human cancer. Despite its important role in the initiation and progression of human cancer, the Normal cells in the bodyare programmed bya variety underlying mechanism of RAS-induced human epithelial of distinct signals to grow, divide, and eventuallydie in a transformation remains elusive. In this study, we probe the coordinated fashion. Cancer arises when cells escape cellular and molecular mechanisms of RAS-mediated normal growth control and fail to die. Considerable transformation, by profiling two human ovarian epithelial knowledge of oncogenesis and tumor development has cell lines. One cell line was immortalized with SV40T/t been gained using cellular and animal cancer models. It antigens and the human catalytic subunit of telomerase is clear that oncogenic transformation is a complex (T29), while the second cell line was transformed with an process that involves multiple steps of genetic and additional oncogenic rasV12 allele (T29H). In total, 32 cellular alterations. During this process, cells destined proteins associated with RAS-mediated transformation for oncogenic transformation progressivelyacquire capa- have been identified using peptide mass fingerprinting. bilities of self-sufficiencyin growth signals, insensitivity These protein targets are involved in several cellular to antigrowth signals, limitless replicative potential, pathways, including metabolism, redox balance, calcium evasion of apoptotic signals, tissue invasion and signaling, apoptosis, and cellular methylation. One such metastasis, and sustained angiogenesis (Hanahan and target, the 40kDa procaspase 4 is significantly upregu- Weinberg, 2000). Although identification of various lated at the protein level in RAS-transformed T29H cells, oncogenes and tumor suppressor genes has greatly related directly to signaling through MEK, but not PI3 advanced our understanding of the steps associated kinase. Cellular caspase 4 activity is, however, suppressed with cancer formation, the underlying cellular signaling in the T29H cells, suggesting that the maturation process networks governed byindividual oncogenes or tumor of caspase 4 is abrogated in RAS-transformed T29H cells. suppressor genes remain poorlyunderstood at the Consistent with this notion, transformed T29H cells were molecular level. One important reason for this major less susceptible to the toxic effects of anti-Fas antibody gap is the lack of appropriate human cancer models and than were immortalized, nontransformed T29 cells, subsequently the ability to systematically analyse these associated with less activation of caspase 4. This study model systems. demonstrates that functional proteomic analysis of a Although primaryrodent cells can be readilytrans- genetically defined cancer model provides a powerful formed bytwo cooperating oncogenes (Land et al., approach toward systematically identifying cellular tar- 1983; Ruley, 1983), oncogenic transformation of human gets associated with oncogenic transformation. primarycells was achieved onlyrecentlybyintroducing Oncogene (2005) 24, 6174–6184. doi:10.1038/sj.onc.1208753; a combination of the SV40 T/t oncogenes, the telomer- published online 6 June 2005 ase catalytic subunit gene (hTERT), and an oncogenic allele of H-RASV12 (Hahn et al., 1999). With the advent Keywords: caspase 4; mass spectrometry; ovarian cancer; of this methodology, malignant transformation has been proteomics; ras; transformation demonstrated in a varietyof human cell lines using the same set of defined genetic elements. These systems include human embryonic kidney cells and foreskin fibroblasts (Hahn et al., 1999), primaryairwayepithelial cells (Lundberg et al., 2002), primarymammary epithelial cells (Elenbaas et al., 2001), mesothelial cells *Correspondence: X Cheng; E-mail: [email protected] (Yu et al., 2001), and astrocytes (Rich et al., 2001). Received 18 November 2004; revised 6 April 2005; accepted 12 April Recently, a genetically defined model for human ovarian 2005; published online 6 June 2005 cancer has also been established using normal human Proteomic analysis of ras-mediated oncogenic transformation T Young et al 6175 ovarian surface epithelial (HOSE) cells (Liu et al., 2004). nude mice. Extracts from each cell line were analysed The observation that a few discrete genetic alterations in triplicate by2-DE. From preliminarystudies using a can transform human cells in vitro supports the notion 3–10 pH gradient, we concluded that the majorityof 2- that there maybe common molecular pathways DE detectable proteins fell within a pH range of 5–7 on associated with the seeminglycomplex and diverse the gel. Therefore, a pH gradient of 4–7 was used in the phenotypes of cancer. Equally important, these model first dimension to obtain maximal resolution of protein systems provide an opportunity to study the complex spots. We also found that the use of 10% tricine–SDS processes of tumorigenesis, especiallythe dissection of polyacrylamide gels, instead of glycine–SDS gradient the functional roles of individual tumor promoter and gels, achieved a better dynamic separation range and suppressor genes in oncogenic transformation of human reproducibility. For each cell line, at least two indepen- cells from different organ sites (Watnick et al., 2003; dent cell lysates were obtained from cultures at different Young et al., 2004). earlypassages and four or more well-resolved gels from The small GTPase RAS has long been known to play six different runs were analysed. Figure 1 shows a significant role in tumor formation and development. representative 2-DE images of T29 and T29H; approxi- RAS mutations are found in 30% of all human cancers mately2200 distinct protein spots were resolved within (Bos, 1989). Although multiple mediators of RAS each gel. signaling, including Raf, PI3 kinase, AF-6, and Ral- The intensityof each protein spot was determined, GDS, have been discovered, the exact role that normalized to the sum of intensities of all spots on the oncogenic RAS plays in the multistep process of gel, and quantified as a percentage volume in each gel oncogenesis is not completelyunderstood and many using Phoretix 2-D analysis software (Nonlinear). Each downstream functions of RAS signaling in cellular individual protein spot was then matched with the transformation remain a mystery. In the current study, identical protein spot from each replicate gel. Data for we applied a proteomics approach to analyse a these matched spots were then averaged over replicate transformation model in which normal HOSE cells are gels for each cell line. The average normalized volume of immortalized bySV40 T/t oncogenes and hTERT, and each spot in the transformed T29H cells was then further transformed through the addition of the compared to that of the matched spot in immortalized constitutivelyactive H-RAS V12. Proteomics provides a T29 cells. In order to determine what would constitute powerful tool for the systematic analysis of alterations significant changes between transformed and untrans- in protein expression and post-translational modifi- formed cells, the intrinsic variance of each protein spot cations, and has been successfullyapplied recentlyfor was determined for each cell line. The average variance identifying tumor markers in bladder squamous cell for individual spots in the replicate gels for the T29 carcinomas (Ostergaard et al., 1997), hepatoma (Yu immortalized cell line was 34%. The average variance in et al., 2000), and melanoma (Bernard et al., 2003). By the transformed T29H cell line was similar at 29% profiling the protein expression differences between two (Figure 2). These intrinsic variances determined the geneticallydefined cell lines from an ovarian cancer basal noise levels of our 2-DE analysis. Based on these model using two-dimensional gel electrophoresis observed values, we selected spots whose average (2-DE), we have identified numerous proteins and normalized volume increased or decreased byat least several specific cellular pathways associated with RAS- 50% between the T29 and T29H cell lines as signifi- mediated oncogenic transformation. Manyof these cantlychanged candidates. We further narrowed the targets are novel downstream molecules that have not field of potential differentiallychanged protein by previouslybeen linked to RAS signaling pathwaysor eliminating spots with greater than 50% variance. transformation. Results from our studynot only Finally, two independent observers examined the spots provide further insights into our understanding of the visuallyto confirm that changes from automated RAS-mediated tumorigenic transformation process but imaging analysis were real and reproducible. The final also offer a novel approach for studying cancer signaling analysis revealed 80 protein spots that differed between globally. immortalized T29 cells and H-RASV12 transformed T29H cells. Of these spots, 56 showed increased levels while 14 showed decreased levels in T29H cells. Furthermore, three spots were present onlyin H-RAS Results transformed T29H cells while seven were absent. 2-DE gel analysis Identification of putative protein targets involved To explore the molecular mechanisms of H-RAS in RAS-mediated oncogenic transformation mediated oncogenic transformation, comparative protein profiling was performed using two well-defined cell Of the 80 protein spots showing significant changes in lines, T29 and T29H, derived from HOSE. While T29 protein levels after oncogenic transformation byRAS, cells, stablytransfected with SV40 T/t antigens and we successfullyidentified 32 of them from 34 protein hTERT, are fullyimmortalized, onlythe addition of spots bypeptide mass fingerprinting. As shown in oncogenic RAS leads to the malignant transformation Table 1, the apparent observed molecular mass and of T29H cells as demonstrated bythe capabilityof isoelectric point for all identified proteins matched very anchorage-independent growth and tumor formation in well with their calculated values.

Oncogene Proteomic analysis of ras-mediated oncogenic transformation T Young et al 6176

Figure 1 Proteomic analysis of T29 and T29H cells using 2-DE gels. Whole-cell lysates (200 mg) from T29 (a) and T29H (b) HOSE were separated on 2-DE gel and visualized bysilver staining as described in ‘Materials and methods’. Arrows indicate identiﬁed protein spots signiﬁcantlyaltered between T29 and T29H cells as listed in Table 1

potential increase in cellular antioxidant capacitythat could protect cells from excessive oxidative stress associated with H-RAS-mediated transformation (Young et al., 2004). In addition, several calcium- binding proteins, including calretinin, calponin 3, and calpain small regulatorysubunit 1, were modulated following H-RAS transformation (Figure 3). These proteins are enzymes involved in calcium homeostasis and signaling. Interestingly, two protein spots (7 and 8) with identical molecular weight and slightlydifferent observed pIs of 5.8 and 6.1 were identified as the same calcium binding protein, acidic calponin 3. Both spots showed a 2- to 2.5-fold increase in protein level in the Figure 2 Statistical analysis of intrinsic variance of protein spots T29H cells. This type of change in calponin 3 pI, with identified in 2-DE gels. Average standard derivations of normalized spot volume of individual protein spot for T29H and T29 cell lines. little or no alteration in molecular mass, suggested The average variance for individual spots in the replicate gels possible post-translational modification, such as phos- (n ¼ 4) for the T29 immortalized cell line was 34%. The average phorylation. variance in the transformed T29H cell line in the replicate gels (n ¼ 4) was similar at 29% Post-translational regulation of caspase 4 maturation by RAS oncogenic transformation Most of these proteins could be classified into five Expression levels of several proteases, including cathe- general groups according to their cellular function, psin D, calpain, and procaspase 4, were altered in HOSE including apoptosis/proteolysis, calcium signaling, cel- cells oncogenicallytransformed byH-RAS. One apop- lular methylation, metabolism, and redox homeostasis tosis related cysteine protease, the B40 kDa proform of (Table 2). The largest protein group consisted of caspase 4, was upregulated about four-fold in T29H enzymes involved in cellular metabolism. A total of 10 cells in our 2-DE gel analysis (Figure 4a). The increased proteins have been identified within this categoryand level in procaspase 4 was further confirmed by most of these proteins were upregulated. In conjunction, immunoblotting analysis using caspase 4 specific anti- five cellular enzymes involved in metabolizing reactive bodies (Figure 4b). Surprisingly, when the cellular oxidative species (ROS) and maintaining redox balance activities of caspase 4 in T29 and T29H cells were in cells were also found to be upregulated. The measured using a caspase 4 specific substrate, we found upregulation of the ROS-related enzymes argued for a that the actual basal cellular caspase 4 activitywas

Oncogene Table 1 Proteins identiﬁed bymass spectrometryto be changed signiﬁcantlybetween T29 and T29H cells a

Spot # Protein ID NCBI Acc # Number peptides MW pI MWObs pIObs Sequence Fold change matched (mean coverage (%) (s.d.) mass error)

1 Copine I isoform 1 NP_003906 13 (0.030) 59.2 5.5 60 5.7 28 À2.76 (0.73) 2 Aldehyde dehydrogenase, mitochondrial precursor P05091 11 (0.017) 56.4 7 55 6.1 28 1.54 (0.20) 3 Mannose-6-phosphate isomerase NP_002426 10 (0.031) 46.6 5.6 46 5.8 34 1.62 (0.21) 4 Caspase 4 isoform alpha precursor NP_001216 8 (0.035) 43.3 5.7 49 6.0 26 4.12 (0.76) 5 Ornithine aminotransferase mitochondrial precursor P04181 5 (0.029) 48.5 6.6 47 6.5 16 À1.66 (0.27) 6 Met Adenosyltransferase isoform 2: subunit alpha NP_005902 13 (0.039) 43.6 6 44 6.8 40 1.53 (0.17) 7 Calponin 3, acidic NP_001830 14 (0.022) 36.4 5.7 40 6.1 40 2.09 (0.37) 8 Calponin 3, acidic NP_001830 10 (0.016) 36.4 5.7 40 5.8 37 2.54 (0.34) 9 Selenophosphate synthetase NP_036379 11 (0.050) 42.9 5.6 44 5.9 31 1.55 (0.20) 10 SPFH domain family, member 2 isoform 1 NP_009106 7 (0.064) 38.1 5.5 42 5.6 28 1.71 (0.36) 11 Capping protein (actin ﬁlament) gelsolin like NP_001738 6 (0.021) 38.5 5.9 40 6.3 24 1.68 (0.26) 12 3-hydroxyisobutyrate dehydrogenase NP_689953 8 (0.021) 32 5.8 35 5.8 20 1.60 (0.22) 13 Aminoadipate-semialdehyde-dehydrogenase-phospho- NP_056238 10 (0.028) 35.8 6.4 38 6.4 33 1.89 (0.29) pantetheinyl-transferase 14 DCI Protein (3,2 trans-enoyl-coenzyme A isomerase) AAH09631 10 (0.045) 29.3 6.4 30 6.5 47 1.73 (0.31) 15 Enoyl-CoA hydratase 1, mitochondrial short chain NP_004083 10 (0.025) 31.8 8.9 29 6.3 45 1.67 (0.17) precursor 16 Thioredoxin peroxidase NP_006397 9 (0.041) 30.5 5.9 28 5.9 53 1.50 (0.20) 17 Replication protein A2 (32 kDa) NP_002937 5 (0.009) 29.2 5.8 32 6.0 27 2.08 (0.28) 18 Quinolinate Phosphoribosyltransferase NP_055113 5 (0.062) 31.1 5.8 32 6.0 20 16.13 (4.97) rtoi nlsso a-eitdocgnctransformation Young oncogenic T ras-mediated of analysis Proteomic 19 6-phosphogluconolactonase NP_036220 13 (0.023) 27.5 5.7 30 6.0 66 1.59 (0.20) 20 Protein-L-Isoaspartate O-methyltransferase, chain A NP_005380 8 (0.024) 24.6 6.7 26 6.5 42 1.99 (0.49)

21 Peroxiredoxin 3, isoform b NP_054817 5 (0.022) 26.1 7.1 23 6.4 21 1.82 (0.21) al et 22 Guanidinoacetate N-methyltransferase NP_000147 6 (0.016) 26.3 5.7 26 5.9 30 3.90 (1.22) 23 NADH dehydrogenase Fe/S protein 3 30 kDa (NADH- NP_004542 8 (0.077) 30.2 7 27 6.0 32 2.20 (0.35) coenzyme Q reductase) 24 Mago Nashi Homolog protein NP_002361 6 (0.033) 17.1 5.7 16 5.9 43 4.77 (2.12) 25a Membrane-type 1 matrix metalloproteinase cytoplasmic NP_060739 9 (0.007) 21.5 5.4 18 5.5 49 À2.20 (0.67) tail binding protein-1 25b Stathmin-1 NP_005554 9 (0.006) 17.3 5.8 18 5.5 43 À2.20 (0.67) 26 Phosphohistidine Phosphatase I NP_054891 6 (0.024) 14 5.7 13 5.5 51 2.03 (0.33) 27 Phosphoserine Phosphatase NP_004568 6 (0.033) 25.0 5.5 25 5.5 36 7.34 (2.35) 28 Latexin NP_064554 6 (0.040) 25.8 5.5 30 5.6 22 2.34 (0.31) 29 Tumor susceptibilityprotein NP_006283 5 (0.039) 43.9 6.1 45 6.2 32 1.83 (0.26) 30 Glyoxalase I NP_006699 10 (0.026) 20.7 5.2 25 5.0 66 2.03 (0.13) 31 Calpain small subunit I NP_001740 11 (0.035) 28.3 5.1 28 4.9 44 À1.54 (0.16) 32 Calretinin (calbindin 2) NP_001731 8 (0.017) 31.5 5.1 32 5.0 32 NA 33 Cathepsin D (31 kDa cleavage product) NP_001900 15 (0.046) 45.1 6.1 30 5.3 35 2.08 (0.54) 34 Cathepsin D (31 kDa cleavage product) NP_001900 11 (0.014) 45.1 6.1 31 5.4 32 À2.96 (0.37)

Shaded portion indicates proteins that have been linked to cancer or RAS signaling. aProtein fold change was calculated using Nonlinear Phoretix 2-D Advanced software as described in Material and methods. Positive and negative numbers indicate increases and decreases in protein levels, respectively, in Ras-transformed T29H cells relative to T29 cells. The spot numbers correspond to those on the master images shown in Figure 1 Oncogene 6177 Proteomic analysis of ras-mediated oncogenic transformation T Young et al 6178 Table 2 Functional groups/pathways of proteins changed signiﬁcantly between T29 and T29H cells Group Protein ID Function

Metabolism Mannose-6-phosphate isomerase Mannose biosynthetic enzyme, necessary for lipid glycosylation Metabolism 3-hydroxyisobutyrate dehydrogenase Enzyme involved in amino-acid catabolism Metabolism Ornithine aminotransferase mitochondrial precursor Mitochondrial enzyme controls L-ornithine levels and ammonia metabolism Metabolism Quinolinate phosphoribosyltransferase Enzyme involved in catabolism of quinolinic acid and NAD+ synthesis Metabolism Aminoadipate-semialdehyde-dehydrogenase-phospho- Enzyme involved in lysine catabolism pantetheinyl-transferase Metabolism DCI Protein (3,2 trans-enoyl-coenzyme A isomerase) Fatty acid oxidation enzyme, in peroxisomes Metabolism Enoyl-CoA hydratase 1, mitochondrial short chain Fattyacid oxidation enzyme, in mitochondria precursor Metabolism 6-phosphogluconolactonase Enzyme involved in pentose phosphate pathway Metabolism Phosphoserine Phosphatase Enzyme involved in the synthesis of L-serine Metabolisma Aldehyde dehydrogenase, mitochondrial precursor Alcohol metabolism enzyme, possible cancer biomarker Redox balance Glyoxalase I Recently found as a biomarker for invasive ovarian cancer Redox balance Selenophosphate synthetase Essential in selenoprotein biosynthesis, helps maintain redox balance, antioxidant Redox balance Thioredoxin peroxidase Mitochondrial antioxidant protein, catalyses free radi- cals Redox balance NADH dehydrogenase ubiquinone Fe/S protein 3 Nuclear encoded FeS protein component of complex I in (30 kDa) mitochondrial respiratorychain Redox balance Peroxiredoxin 3, isoform b Mitochondrial antioxidant protein, antiapoptotic effects Ca2+ signaling Copine I isoform 1 Ca2+ dependent phospholipid binding protein Ca2+ signaling Calponin isoform 3, acidic Smooth muscle associated protein, involved in Ca2+ dependent muscle movement and cellular actin rearran- gement Ca2+ signaling Calretinin (calbindin 2) Biomarker for certain cancers, Ca2+ binding protein involved in Ca2+ homeostasis Ca2+ signalingb Calpain small subunit I Ca2+ dependent protease involved in Ca2+ signaling Protease Caspase 4 isoform alpha precursor Apoptosis related cysteine protease Protease Cathepsin D (31 kDa cleavage product) Lysosomal protease involved with invasiveness and progression, also mediates apoptosis Methylation Met Adenosyltransferase isoform 2: subunit alpha Catalyses formation of S-adenosyl-methionine, activity in growth and differentiation in some cells Methylation Protein-L-Isoaspartate O-Methyltransferase, chain A Involved in protein repair, protein methylation, and possiblycell motility Methylation Guanidinoacetate N-methyltransferase Methylates guanidinoacetate, important in creatine biosynthesis

The shaded portion indicates different groups of protein based on their functions. aAldehyde dehydrogenase 2 can be classiﬁed as both a metabolic protein and a reducer of oxidative stress. bCalpain is a calcium-dependent protease involved in apoptotic cascades

Figure 3 Identiﬁed cellular calcium binding proteins signiﬁcantlyaltered in T29H cells. While protein levels of calponin 3 were increased for 2.5 (left arrow) and 2.0 (right arrow) folds, the levels of calpain 1 and copine 1 were decreased for 1.5- and 2.8-folds in T29H cells, respectively. The calretinin spot was detected only in gels for T29H cells

Oncogene Proteomic analysis of ras-mediated oncogenic transformation T Young et al 6179 activation is directlyresponsible for suppressing the processing of procaspase 4 in T29H cells, the effect of selectivelysilencing cellular RAS on the levels of procaspase 4 in T29H cells was examined. Expression of two retroviral-mediated siRNA vectors, H1-siRNA, which targets the H-RASV12 mutation, and H2-siRNA, which targets sequences in the wild-type H-RAS, significantlyreduced the levels of procaspase 4 (Figure 4d) to that of T29. To further confirm that activation of RAS pathways was important for suppressing the maturation and activation of caspase 4 in T29H cells, we monitored the levels of procaspase 4 in T29H cells in response to various specific inhibitors that target the RAS downstream effectors MEK and PI3K. Inhibition of MEK byU0126 led to a significant reduction of procaspase 4 protein level while LY294002, a PI3 Kinase specific inhibitor, had little effect on the level of procaspase 4 (Figure 4e). These results clearlydemonstrated that activation of RAS and its downstream MAP kinase pathwaywas directly responsible for the suppression of caspase 4 processing and maturation in T29H cells.

RAS suppresses caspase 4 maturation and Fas-mediated apoptosis Caspase 4 activation has been implicated in Fas- mediated apoptosis (Kamada et al., 1997; Martin and Panja, 2002). The apparent increased level of procaspase 4 protein and decreased caspase 4 activityin T29H cells led us to hypothesize that oncogenic RAS may be protecting the transformed T29H cells from Fas- mediated apoptosis bysuppressing the maturation of caspase 4 and/or other related proteases involved in cellular apoptosis. To test this hypothesis, we subjected the T29 and T29H cells to an agonist anti-Fas antibody CH11 (Upstate Biotechnology) that crosslinks the death receptor Fas and leads to the formation of a death- inducing signaling complex. As shown in Figure 5, treatment of CH11 antibodyled to massive cell death Figure 4 Post-translational regulation of procaspase 4 maturation and significant apoptosis indicated bythe activation of byRAS oncogene. ( a) Protein levels of procaspase 4 in T29H and caspase 3 in T29 cells, while the transformed T29H cells T29 cells revealed by2-DE gel analysis.( b) Protein levels of were not significantlyaffected byCH11 treatments procaspase 4 in T29H and T29 cells revealed byimmunoblotting under the same conditions. When cell lysates from T29 analysis using caspase 4 specific antibodies. (c) Cellular caspase 4 activities in T29 and T29H cells measured using caspase 4 specific and T29H cells treated with CH11 antibodywere substrate. (d) Protein levels of procaspase 4 in T29, T29H, and probed with a caspapse 4 specific antibody, activation T29H with cellular RAS protein silenced bysiRNA (H1 and H2) of caspase 4 as monitored bythe formation of active revealed byimmunoblotting analysisusing caspase 4 specific caspase 4 P20 and P10 subunits was observed in T29 antibodies. (e) Protein levels of procaspase 4 in T29, T29H, and cells, while significantlyless P20 and P10 subunits and a T29H treated with MEK specific inhibitor U0126 and PI3K specific inhibitor LY294006 revealed byimmunoblotting analysisusing concomitant increase in caspase 4 intermediates were caspase 4 specific antibodies detected in T29H cells (Figure 5c). These results further confirmed that RAS mediated oncogenic transformation blocked the maturation of caspase 4 and attenuated decreased in T29H cells (Figure 4c). Since the active Fas-mediated apoptosis in HOSE cells under the state of caspase 4 consists of a tetramer of P20 and P10 conditions described. subunits derived from the proteolytic processing of procaspase 4, this decrease in enzymatic activity coupled with an apparent increase in the proform of caspase 4 Discussion protein in T29H cells suggested that maturation and activation of this enzyme were blocked in T29H cells Extensive studies have been conducted to discern the following H-RAS transformation. To determine if RAS signaling pathways activated by the oncogene RAS in

Oncogene Proteomic analysis of ras-mediated oncogenic transformation T Young et al 6180 the last 20 years. RAS mutations have been found in about 30% of all human cancers, and manylines of evidence support a primaryrole for RAS signaling pathways in ovarian cancer specifically. Mutations in K-RAS and its downstream effector B-RAF are frequently detected in serous borderline and low-grade serous carcinoma (Singer et al., 2003; Sieben et al., 2004). Activating mutations of H-RAS are present in about 6% of ovarian cancers (Varras et al., 1999). Moreover, activation of H-RAS upstream and downstream effector pathways is often demonstrated in the absence of H-RAS mutation (Patton et al., 1998). For example, the upstream signaling molecule Her-2/Neu and immediate downstream RAS effector B-RAF were found to be upregulated and active in a large portion of ovarian cancers (Berchuck et al., 1990; Gemignani et al., 2003). Most importantly, the RAS-transformed human ovarian cancer model used in this studyexhibits many similar features of natural ovarian tumors. Gene expression profile analysis reveals that both RAS- transformed cells and naturallyderived ovarian cancer cells use the NF-kB pathwayto activate cytokines and facilitate tumor formation (Liu et al., 2004). Although it is clear that RAS plays an important role in tumorigenesis, the molecular mechanism of RAS-mediated oncogenic transformation as a whole has not yet been identified. We chose a proteomics approach to system- aticallyanalyse the alterations in the cellular proteome of this model system of human ovarian epithelial cells transformed byH-RAS. The use of a geneticallydefined cancer transformation model allows the dissection of downstream signaling pathways specifically associated with RAS-mediated oncogenic transformation, while our proteomic analysis enabled the identification of numerous changes in protein expression as well as possible post-translational processing and proteolytic cleavage that are not amenable to genomic analysis. This is underscored bythe fact that onlyabout 15% of all target proteins identified bythe proteomic approach in this studyshowed concomitant changes in mRNA abundance as detected bycDNA expression array(Liu et al., 2004). Proteins identified in our 2-DE proteomic screen of H-RAS transformed cells included those that are involved in multiple cellular processes such as metabolism, redox regulation, calcium signaling, apoptosis, and cellular methylation (Table 2). The largest group of proteins identified consisted of enzymes involved with metabolic processes. This was not surprising since tumor Figure 5 Resistance to Fas-mediated cell killing in T29H cells. (a) T29 and T29H cells treated with 100 ng of anti-Fas monoclonal cells often require higher metabolic levels to maintain a antibodyCH11 for 96 h and observed under a phase-contrast growth advantage over normal cells and was consistent microscope. (b) Cellular caspase 3 activityof T29 and T29H cells with the notion of increased energydemands in cancer with or without the treatment of various concentrations of CH11 cells (Argiles and Azcon-Bieto, 1988). Several of these for 96 h. (A–D) T29 cells treated with 0, 25, 100, and 200 ng/ml of metabolic enzymes, including aldehyde dehydrogenase CH11 antibody, respectively. (E–H) Data from T29H cells treated with the corresponding concentrations of CH11 at 0, 25, 100, and 2, ornithine aminotransferase, and enoyl-CoA hydratase 200 ng/ml. Data are normalized against T29 untreated cells. Error were previouslyimplicated in tumor formation (Mat- bars indicate standard deviation (n ¼ 3). (c) Processing of thaei and Williams, 1987; Lindahl, 1992; Balabanov procaspase 4 into active P20 and P10 caspase 4 subunits in T29 et al., 2001). Interestingly, several proteins involved in and T29H cells treated with 300 ng/ml of CH11 for 96 h revealed by immunoblotting analysis using caspase 4 specific antibodies metabolizing reactive oxygen species and maintaining redox balance were also upregulated in T29H cells. Upregulation of these antioxidant proteins allows T29H

Oncogene Proteomic analysis of ras-mediated oncogenic transformation T Young et al 6181 cells to offset high levels of reactive oxygen species that cleavage of procaspase 4 into a smaller catalytically are associated with increased metabolism. This increase active enzyme, and given the fact that while levels of the in antioxidant proteins confers resistance to ROS procaspase 4 were increased, overall basal cellular mediated apoptosis in RAS transformed cells (Young caspase 4 activitywas actuallydecreased (Figure 4c), et al., 2004). we concluded that the maturation of procaspase 4 was Manyof the identified protein targets, including blocked at the post-translational level in RAS-trans- calretinin, cathepsin D, glyoxylase I, aldehyde dehydro- formed T29H cells. The fact that upregulation of genase 2, and the calcium dependent protease Calpain 1, procaspase 4 could be reversed byU0126, an MEK are associated with cancer development or RAS signal- specific inhibitor, demonstrated that this inhibitory ing. Calretinin, a mediator of intracellular calcium effect on caspase 4 maturation was directlymediated responses (Rogers et al., 1990), is a biomarker used in bythe MAP kinase pathwaydownstream of RAS. This diagnoses of cancers, including epithelial ovarian inhibitorymechanism could be due to a phosphoryla- tumors (Cao et al., 2001). Cathepsin D has been tion modification of procaspase 4 byMAP kinase, in a implicated in numerous studies as a cysteine protease manner similar to procaspase 9 whose maturation has involved in extracellular proteolysis of the basement been shown to be blocked byH-RAS activation via membrane and metastasis to foreign tissue. Upregula- phosphorylation at Thr125 (Allan et al., 2003). The tion of cathepsin D is associated with poor prognosis in exact mechanism of RAS-mediated inhibition of caspase breast and ovarian cancer patients (Losch et al., 2004). 4 activation is currentlyunder further investigation. Transformation byoncogenic RAS has been shown to Caspase 4 has been implicated in the Fas-mediated alter the processing and intracellular trafficking of apoptotic cascade (Kamada et al., 1997; Martin and lysosomal cathepsin D (Demoz et al., 1999) and to Panja, 2002) and decreased sensitivityto Fas-mediated upregulate cathepsin D activityin human breast apoptosis has been shown to contribute to ovarian epithelial cell lines (Calaf et al., 1994). In addition, tumorigenesis. The apparent blockage of caspase 4 glyoxylase I has been recently identified as a biomarker activation and resistance to Fas-mediated apoptosis for ovarian cancer (Jones et al., 2002), while the suggest that RAS-mediated suppression of caspase 4 expression of aldehyde dehydrogenase 2 is altered activation mayallow H-RAS transformed cells to evade significantlyin hepatocellular carcinomas (Park et al., apoptosis. 2002). The 28 kDa calpain subunit is the regulatoryarm In conclusion, this studydemonstrates the value of of the calpain system. This subunit binds to calcium and applying a proteomics approach to identify novel modulates the activityof the calpain protease (Goll effectors of H-RAS mediated transformation in human et al., 2003) The calpain system is known to mediate ovarian epithelial cells. Using this method, we have cellular signaling pathways associated with cell prolif- identified numerous proteins that are altered in associa- eration, differentiation and death in response to tion with RAS-mediated oncogenic transformation. The increases in intracellular Ca2 þ . For example, calpain is proteins identified demonstrate that H-RAS mediated implicated as a downstream effector of the Gq signaling transformation of ovarian epithelial cells affects many pathwayfor inhibition of Wnt/ b-catenin-regulated cell aspects of cellular functions, including the loss of proliferation (Li and Iyengar, 2002). Calpain has also sensitivityto apoptosis, and thus confers an aggressive been reported to be important for ERK/MAP kinase oncogenic phenotype on cancer cells. activation associated with epidermal growth factor receptor-mediated fibroblast motility(Glading et al., 2000) and for integrin-induced signaling upstream of Materials and methods Rho GTPases (Kulkarni et al., 1999). In addition, calpain is involved in caspase-independent apoptosis in Cell lines and culture methods ovarian and breast cancers (Bao et al., 2002). The alteration of these known cancer associated proteins and HOSE cells were immortalized and transformed as previously described (Liu et al., 2004). Briefly, isolated human surface biomarkers further supports our H-RAS mediated ovarian epithelium cells were infected sequentiallybyretro- transformation model as a mimic of naturallyoccurring viruses containing SV40 T/t antigens and hTERT genes to cancers in vivo. In addition, the abilityto identifythese generate T29 cells. The immortalized but nononcogenic T29 biomarkers and cancer-related proteins also validates cells were further transformed byintroducing an oncogenic our proteomic approach in studying the mechanism of H-RASV12 in a pLNCX retroviral vector to form the T29H cell oncogenic transformation. line. Cells were cultured in MCDB105/Media 199 medium Of the proteins identified, onlyabout 50% had a (1 : 1) containing 10% FBS and 1% penicillin/streptomycin. connection with H-RAS signaling or oncogenic trans- Immortalized cells containing the hTERT gene were selected using hygromycin (100 mg/ml) and transformed cells contain- formation established in the literature. The remainder V12 consisted of proteins with no known association with ing both hTERT and H-RAS were selected with hygromycin and puromycin (1 mg/ml). Earlypassages of the T29 and T29H RAS signaling or transformation. These proteins pre- cells were used. sent a unique opportunityto delineate new pathways that might be involved in the transformation process mediated byH-RAS. For example, we identified an Preparation of total cell lysate apparent upregulation of procaspase 4 in the RAS- To obtain total protein lysates, 70–80% confluent T29 and transformed cells. Since caspase 4 activitydepends on T29H cells from 100 mm plates were trypsinized, pelleted, and

Oncogene Proteomic analysis of ras-mediated oncogenic transformation T Young et al 6182 suspended in 500–1000 ml lysis buffer (7 M urea, 2 M thiourea, Tryptic in-gel digestion and mass spectrometry 4% Chaps, 1 mM EDTA, 1 mM EGTA, 60 mM DTT, 1 mM Protein spots were punched out from multiple wet preparative PMSF, 1 mM benzamidine, 25 mg/ml leupeptin, 10 mg/ml gels using clean pipette tips with their ends cut to match the aprotinin, 10 mM microcystin, 10 mM sodium orthovanadate). size of the spot. Gel pieces from 1 to 4 or 8 to 10 gels were Cells were lysed at room temperature for 1 h with occasional pooled for high and low abundance spots, respectively. Excised mixing and centrifuged at 180 000 g for 1 h at 221C to pellet gel plugs were washed with 100 mM NH CO and 50% insoluble cellular debris. Protein concentration of total cell 4 3 acetonitrile (ACN)/50 mM NH CO each for 20 min and lysate was determined using the Bio-Rad RC/DC Bradford 4 3 subsequentlydehydratedin 50 ml of ACN for 10 min. The assaythat is compatible with detergents and reducing agents. dried gel pieces were rehydrated in 10 ml of sequencing grade Aliquots were loaded immediatelyonto gels or frozen in a dry trypsin (Promega) in 25 mM NH CO to a final concentration ice/methanol bath and stored at –801C until use. 4 3 of 20 mg/ml trypsin and digested overnight at 371C. Super- natant containing tryptic fragments was transferred to a clean 2-DE gel electrophoresis tube and gel pieces were extracted twice with 30 ml of 60% ACN/0.1% trifluoroacetic acid for 20 min in a sonicating In total, 200 mg of total protein was used in analytical gels for water bath. Peptides mixtures were pooled, dried down in a image analysis and spot identification, while 500 mg of total SpeedVac, and resuspended in 5 ml of 0.1% formic acid. protein was used in preparative gels for protein identification. Digested peptide samples were cocrystallized with an equal Samples in 350 ml of lysis buffer containing 0.5% IPG buffer volume of a-cyano-4-hydroxy-trans-cinnamic acid matrix (pH 4-7) was applied onto Immobilin drystrips (pH 4–7, (Hewlett Packard) on a gold-coated sample plate and analysed 18 cm, Pharmacia Biosciences) and allowed to rehydrate at byMALDI-TOF (matrix-assisted laser desorption/ionization 1 1 20 C for 12 h. Isoelectric focusing was performed at 20 Con time-of-flight) mass spectrometry(Voyager-DESTR, Applied an IPGphor unit (Pharmacia Biosciences) for 1 h at 500 V, 1 h Biosystems) to obtain peptide masses. Data were summed over at 1000 V, followed bya gradient from 1000 to 8000 V in 1 h 150 acquisitions in delayed extraction mode, with sensitivity and 6 h at 8000 V. After focusing, strips were either immedi- B10 fmol (20 kV accelerating voltage, 10 V guide wire voltage, ately loaded onto polyacrylamide gels or frozen at –801C. 2 100 ns delay). Internal calibrations were performed using Large format (20 Â 20 cm ) 10% tricine–SDS gels were poured tryptic autolysis peaks at 842.5090 and 2211.1064 Da. Peptide fresh the dayof use. Before the second dimension, strips were mass data were analysed by searching against the National incubated in reducing buffer (6 M urea, 2% SDS, 50 mM Tris, Center for BiotechnologyInformation (NCBI) nonredundant 30% glycerol, 65 mM DTT, pH 6.8) for 15 min with shaking at database using Protein Prospector (http://prospector.ucsf.edu/ room temperature, followed by15-min incubation in alkyla- , Universityof California, San Francisco) and Profound tion buffer (6 M urea, 2% SDS, 0.375 M Tris, 30% glycerol, (http://prowl.rockefeller.edu/cgi-bin/ProFound, Rockefeller 240 mM iodoacetamide, pH 8.8), and then loaded onto the University), with mass tolerance of 0.13 Da and allowing one second-dimension gel, sealed with 0.5% low-melting agarose in missed cleavage and peptide modifications byacrylamide running buffer at the top. The second dimension was adducts with cysteine and methionine oxidation. Proteins performed on a Bio-Rad Protean IIxi 2D unit at a constant identified bypeptide mass fingerprinting were further eval- voltage of 140 for 22 h. During the entire course of the uated bycomparing the calculated and observed molecular electrophoresis, the temperature of the gels was maintained at mass and pI, as well as the number of peptides matched and 1 a constant 12 C bya circulating water bath to maximize percent sequence coverage. separation resolution. Analytical gels used for image analysis were stained as described previously(Blum et al., 1987). Preparative gels used for identification of proteins were stained Immunoblotting analysis using a low fixation silver protocol (Shevchenko et al., 1996). Protein concentration of cell lysates was assayed with the Bio- Rad protein assayreagent. Equal amounts of protein (30 mg) Image analysis were loaded onto 12% SDS polyacrylamide mini-gels (Bio- Rad), transferred to PVDF membranes. PVDF blots and the Digital gel images were obtained using a CCD camara (Alpha remaining polyacrylamide gels were stained with Ponceau S Innotech imager 5500). Gel images were then analysed using and Coomassie Blue, respectively, to ensure equal loading and Phoretix 2-D Advanced software (Nonlinear). This software even transfer of the samples. After being blocked overnight in identifies protein spots within each image and matches 5% milk in TBS-Tween, blots were incubated with Anti- identical protein spots across all images. A total of six gels Caspase-4 P-20 antibody(1 : 1000, Santa Cruz) for 1.5 h, from at least two independent cell lysate preparations were followed byHRP-conjugated secondaryantibody(1 : 4000, used from each cell line. Once spots were matched, images Chemicon) for 45 min. Antigen–antibodycomplexes were were manuallyedited to confirm proper spot detection and detected byenhanced chemiluminescence (Pierce). matching. The intensityof each protein spot was normalized as a percentage of total volume, corresponding to pixel intensity Caspase activity assays integrated over the area of each spot and divided bythe sum of all spots in the gel to account for staining variability. T29 and T29H cells were grown in 100 mm plates to 70–80% Following manual editing and matching confirmation, average confluence, then harvested bytrypsinization,washed twice in normalized spot volumes (pixel intensityover spot area) were PBS and once in wash buffer (100 mM HEPES pH 7.4, 0.5 mM compared between T29 and T29H cells. Target candidates EDTA, 1 mM DTT, protease inhibitors 0.1 mM PMSF, 2 mg/ml were identified as protein spots that changed at least 1.5-fold leupeptin, 2 mg/ml pepstatin). The cells were lysed on ice for between immortalized (T29) and transformed (T29H) cell lines 30 min with frequent vortexing in 150 ml of lysis buffer (wash and that were either present or absent in one cell line or the buffer plus 1% Triton X-100). Cell debris was removed by other. Protein spots with greater than 50% internal variance centrifugation at 16 000 g for 20 min at 41C. Lysate protein were removed from the target list. Finally, remaining concentration was determined byBradford assay(Bio-Rad). individual candidates were visuallyexamined to ensure that Caspase 3 and 4 activities were measured using fluorogenic the change was consistent in all gels. substrates DEVD-AFC and Ac-LEVD-AFC that are specific

Oncogene Proteomic analysis of ras-mediated oncogenic transformation T Young et al 6183 for caspases 3 and 4, respectively, by monitoring the release of serum that had been inactivated at 561C for 30 min. After 96 h, free AFC as determined using a Fluoroskan fluorescence cells were examined under phase contrast microscope. For microplate reader with excitation and emission set to 405 and determination of Caspase 3 activity, cells were grown in 10 cm 510 nm, correspondingly. culture plates for 24 h and treated with 0, 25, 100, and 200 ng/ ml CH11 for 96 h and subsequentlyharvested for caspase 3 Inhibition of H-RAS gene expression by siRNA activitymeasurements. T29H cells were transfected with either one of the two retrovirus-mediated H-RAS siRNA vectors (designated as Abbreviations H1 and H2) that have been described previously(Yang et al., V12 2-DE, two-dimensional electrophoresis; ACN, acetonitrile; 2003). Briefly, H1 selectively silences mutant H-RAS while HOSE, human ovarian surface epithelia cells; hTERT, human H2 suppresses both mutant and wild-type H-RAS expression. telomerase catalytic subunit; MALDI-TOF, matrix-assisted T29H cells were infected with H1 and H2 retrovirus generated laser desorption/ionization time-of-flight. from Phoenix cells and selected for 7–10 days in 0.7 mg/ml of G418 to establish stable cell lines. The expression and activation levels of H-RAS proteins in these siRNA stable Acknowledgements cell lines were assayed by immunoblotting analysis and RAS We are indebted to Dr Natalie Ahn (Universityof Colorado, activityassayusing GST-Raf-RBD (Young et al., 2004), Boulder) for her continuous support and insightful discus- respectively. sions. We thank Dr TonyHaag in the Biomedical Resource Facility(supported byNCI grant R24CA88317, NIEHS Center Grant ES06676 and N01-HV-28184) for mass spectro- CH11 antibody treatment metryanalysis.This work is supported byAmerican Cancer T29 and T29H cells were plated at 1 Â 104 cells/well in 12-well SocietyResearch Scholar Grant RSG-01-035-01-TBE and plates and grown for 24 h. Cells were then treated with CH11 National Institute of Health Grant GM060170 to XC. JL is anti-Fas antibody(Upstate Biotechnology)at concentrations supported byan American Cancer SocietyResearch Scholar ranging from 25 to 300 ng/ml in medium with 10% fetal bovine Grant RSG-04-028-01-CCE.

References

Allan LA, Morrice N, BradyS, Magee G, Pathak S and Clarke Jones MB, Krutzsch H, Shu H, Zhao Y, Liotta LA, Kohn EC PR. (2003). Nat. Cell Biol., 5, 647–654. and Petricoin III EF. (2002). Proteomics, 2, 76–84. Argiles JM and Azcon-Bieto J. (1988). Mol. Cell Biochem., 81, Kamada S, Washida M, Hasegawa J, Kusano H, Funahashi Y 3–17. and Tsujimoto Y. (1997). Oncogene, 15, 285–290. Balabanov S, Zimmermann U, Protzel C, Scharf C, Klebingat Kulkarni S, Saido TC, Suzuki K and Fox JE. (1999). J. Biol. KJ and Walther R. (2001). Eur. J. Biochem., 268, 5977–5980. Chem., 274, 21265–21275. Bao JJ, Le XF, Wang RY, Yuan J, Wang L, Atkinson EN, Land H, Parada LF and Weinberg RA. (1983). Nature, 304, LaPushin R, Andreeff M, Fang B, Yu Y and Bast Jr RC. 596–602. (2002). Cancer Res., 62, 7264–7272. Li G and Iyengar R. (2002). Proc. Natl. Acad. Sci. USA, 99, Berchuck A, Kamel A, Whitaker R, Kerns B, Olt G, Kinney 13254–13259. R, Soper JT, Dodge R, Clarke-Pearson DL and Marks P. Lindahl R. (1992). Crit Rev. Biochem. Mol. Biol., 27, 283–335. (1990). Cancer Res., 50, 4087–4091. Liu J, Yang G, Thompson-Lanza JA, Glassman A, Hayes K, Bernard K, Litman E, Fitzpatrick JL, Shellman YG, Argast G, Patterson A, Marquez RT, Auersperg N, Yu Y, Hahn WC, Polvinen K, Everett AD, Fukasawa K, Norris DA, Ahn NG Mills GB and Bast Jr RC. (2004). Cancer Res., 64, and Resing KA. (2003). Cancer Res., 63, 6716–6725. 1655–1663. Blum H, Beier H and Gross HL. (1987). Electrophoresis, 8, Losch A, Schindl M, Kohlberger P, LahodnyJ, Breitenecker G, 93–99. Horvat R and Birner P. (2004). Gynecol. Oncol., 92, 545–552. Bos JL. (1989). Cancer Res., 49, 4682–4689. Lundberg AS, Randell SH, Stewart SA, Elenbaas B, Hartwell Calaf G, Tahin Q, Alvarado ME, Estrada S, Cox T and Russo KA, Brooks MW, Fleming MD, Olsen JC, Miller SW, J. (1994). Breast Cancer Res. Treat., 29, 169–177. Weinberg RA and Hahn WC. (2002). Oncogene, 21, Cao QJ, Jones JG and Li M. (2001). Int. J. Gynecol. Pathol., 4577–4586. 20, 346–352. Martin CA and Panja A. (2002). Am. J. Physiol Gastrointest. Demoz M, Castino R, Dragonetti A, Raiteri E, Baccino FM Liver Physiol., 282, G92–G104. and Isidoro C. (1999). J. Cell Biochem., 73, 370–378. Matthaei KI and Williams JF. (1987). J. Natl. Cancer Inst., 79, Elenbaas B, Spirio L, Koerner F, Fleming MD, Zimonjic DB, 805–809. Donaher JL, Popescu NC, Hahn WC and Weinberg RA. Ostergaard M, Rasmussen HH, Nielsen HV, Vorum H, Orntoft (2001). Genes Dev., 15, 50–65. TF, Wolf H and Celis JE. (1997). Cancer Res., 57, 4111–4117. Gemignani ML, Schlaerth AC, BogomolniyF, Barakat RR, Park KS, Cho SY, Kim H and Paik YK. (2002). Int. J. Cancer, Lin O, Soslow R, Venkatraman E and Boyd J. (2003). 97, 261–265. Gynecol. Oncol., 90, 378–381. Patton SE, Martin ML, Nelsen LL, Fang X, Mills GB, Bast RC, Glading A, Chang P, Lauffenburger DA and Wells A. (2000). Jr and Ostrowski MC. (1998). Cancer Res., 58, 2253–2259. J. Biol. Chem., 275, 2390–2398. Rich JN, Guo C, McLendon RE, Bigner DD, Wang XF and Goll DE, Thompson VF, Li H, Wei W and Cong J. (2003). Counter CM. (2001). Cancer Res., 61, 3556–3560. Physiol Rev., 83, 731–801. Rogers J, Khan M and Ellis J. (1990). Adv. Exp. Med. Biol., Hahn WC, Counter CM, Lundberg AS, Beijersbergen RL, 269, 195–203. Brooks MW and Weinberg RA. (1999). Nature, 400, RuleyHE. (1983). Nature, 304, 602–606. 464–468. Shevchenko A, Wilm M, Vorm O and Mann M. (1996). Anal. Hanahan D and Weinberg RA. (2000). Cell, 100, 57–70. Chem., 68, 850–858.

Oncogene Proteomic analysis of ras-mediated oncogenic transformation T Young et al 6184 Sieben NL, Macropoulos P, Roemen GM, Kolkman-Uljee Watnick RS, Cheng YN, Rangarajan A, Ince TA and SM, Jan FG, Houmadi R, Diss T, Warren B, Al Adnani M, Weinberg RA. (2003). Cancer Cell, 3, 219–231. De Goeij AP, Krausz T and Flanagan AM. (2004). Yang G, Thompson JA, Fang B and Liu J. (2003). Oncogene, J. Pathol., 202, 336–340. 22, 5694–5701. Singer G, Oldt R, III, Cohen Y, Wang BG, SidranskyD, Young TW, Mei FC, Yang G, Thompson-Lanza JA, Liu J and Kurman RJ and Shih I. (2003). J. Natl. Cancer Inst., 95, Cheng X. (2004). Cancer Res., 64, 4577–4584. 484–486. Yu J, Boyapati A and Rundell K. (2001). Virology, 290, Varras MN, Sourvinos G, Diakomanolis E, Koumantakis E, 192–198. Flouris GA, Lekka-Katsouli J, Michalas S and Spandidos Yu LR, Zeng R, Shao XX, Wang N, Xu YH and Xia QC. DA. (1999). Oncology, 56, 89–96. (2000). Electrophoresis, 21, 3058–3068.

Oncogene