Comparative Transcriptome and Proteome Analysis of Ha-ras and B-raf Mutated Mouse Liver Tumors Benjamin Rignall,† Carina Ittrich,‡ Eberhard Krause,⊥ Klaus E. Appel,§ Albrecht Buchmann,† and Michael Schwarz*,† Department of Toxicology, Institute of Experimental and Clinical Pharmacology and Toxicology, University of Tu ¨ bingen, Wilhelmstr. 56, 72074 Tu ¨ bingen, Germany, German Cancer Research Center, Central Unit of Biostatistics, Im Neuenheimer Feld 280, 69120 Heidelberg, Germany, Leibniz Institute for Molecular Pharmacology (FMP), Robert-Ro¨ssle-Strasse 10, 13125 Berlin, Germany, and Federal Institute for Risk Assessment, Center for Experimental Toxicology, Thielallee 88-92, 14195 Berlin, Germany Received March 30, 2009
Mouse liver tumors frequently harbor activating mutations in the Ha-ras protooncogene. In addition, mutations are also found in the B-raf gene leading to constitutive activation of the B-Raf kinase. In two previous studies, we have investigated by microarray analysis the effect of the mutations on the mRNA expression patterns of the respective tumors. In the present study, we analyzed proteome changes in Ha-ras and B-raf mutated liver tumors by 2-D gel-electrophoretic separation of proteins followed by their identification by mass spectrometry. In total, 104 significantly altered protein spots were identified in Ha-ras mutated tumors and 111 in B-raf mutated tumors when compared to the corresponding normal liver tissue. The changes in protein expression patterns were highly correlated between Ha-ras and B-raf mutated tumors, and in the majority of the cases, both tumor types showed the respective alteration. Most of the tumor-specific changes in protein expression were reflected by similar changes in their mRNAs except for some up-regulated proteins without accompanying changes in mRNA levels. Interestingly, Ha-ras but not B-raf mutated tumors showed high levels of the phosphorylated (activated) form of the Ras/Raf/MEK effector kinase ERK which was, however, not associated with any detectable difference in the transcriptome or protein setup of the tumors. Keywords: Ha-ras • B-raf • hepatocarcinogenesis • mouse liver tumors • proteomics • microarray analysis
Introduction Chemically induced mouse liver tumors are frequently mutated in Ha-ras, B-raf or Ctnnb1, leading to constitutive activation of signaling pathways downstream of the respective regulatory proteins affected by mutation of their genes. While Ctnnb1 mutations are highly prevalent in tumors induced by a treatment regiment including phenobarbital or a phenobarbital-like liver tumor promoter,1-3 Ha-ras or, alternatively, B-raf mutations are seen in up to 70% of spontaneously occurring liver tumors or in tumors induced by exclusive treatment with a genotoxic liver carcinogen like N-nitrosodiethylamine.4-6 The B-Raf kinase is downstream of the small monomeric G-protein Ha-Ras and is thought to signal primarily through the MEK/ ERK (mitogen activated ERK kinase/extracellular signal regulated kinase) pathway (reviewed in refs 7-9). Ras proteins, however, can also activate signaling pathways other than MEK/ ERK and thus have a broader spectrum of downstream effectors * Corresponding author: Prof. Dr. Michael Schwarz, Institute of Pharmacology and Toxicology, Department of Toxicology, University of Tu ¨ bingen, Wilhelmstr. 56, 72074 Tu ¨ bingen, Germany, Tel.: +49-7071-29-77398. Fax: +49-7071-29-2273. E-mail:
[email protected]. † University of Tu ¨ bingen. ‡ German Cancer Research Center. ⊥ Leibniz Institute for Molecular Pharmacology (FMP). § Federal Institute for Risk Assessment. 10.1021/pr9002933 CCC: $40.75
2009 American Chemical Society
than Raf kinases; for reviews see Malumbres and Barbacid10 and Downward.11 One would therefore expect that activation of Ha-Ras in mouse liver tumors would impact on the activity of a wider range of transcriptional regulators and therefore produce a transcriptional signature which is different from B-raf mutated tumors. However, the results of a recent microarray analysis conducted in our laboratory did not reveal significant transcriptome differences between Ha-ras and B-raf mutated tumors.12 This unexpected finding suggests that the transcriptional regulators activated by mutated Ha-Ras and B-Raf are largely identical. It is well-documented that the expression of certain proteins is regulated primarily at the transcriptional level, whereas the expression of numerous others is further controlled by complicated post-transcriptional mechanisms including alterations in mRNA stability, translation kinetics, or post-translational modifications affecting protein stability and enzymatic activity. In fact, a number of publications aiming at integrating proteomic and transcriptomic data in yeast and mouse models as well as in human tissues have been recently published often showing considerable discrepancies between mRNA and protein expression patterns.13-19 This has prompted us to comparatively analyze the proteome of Ha-ras and B-raf mutated tumors, aiming to discriminate between the two tumor genoJournal of Proteome Research 2009, 8, 3987–3994 3987 Published on Web 05/28/2009
research articles types based on their protein expression patterns. In addition, we compared changes in the transcriptome of Ha-ras and B-raf mutated liver tumors as detected in earlier studies from our group12,20 with the proteome data from our present experimental work.
Methods Tumor Material. The tumors analyzed in this study were available to us from an experiment conducted earlier in our laboratory.21 In short, 2 weeks old male C3H mice were injected a single intraperitoneal dose of N-nitrosodiethylamine (10 µg/g of body weight). Liver tumors were isolated 35 weeks after carcinogen treatment. Aliquots of liver tumors and surrounding normal liver tissue were shock-frozen in liquid nitrogen and subsequently stored at -70 °C. The analysis for mutations in the Ha-ras or B-raf oncogenesaimed to compare the mutation status with the proteome pattern of the tumorsswas carried out as described earlier by restriction length polymorphism analysis of PCR-amplified DNAs from tumors and normal liver using restriction enzymes that are diagnostic for the known hotspot mutations in the two genes.5,6 Analysis for mutations in tumors present in stained liver sections was essentially done as previously described.6,22 In brief, paraffin liver sections were prepared, stained with hemalaun/eosin and mounted on dialysis bags. Thereafter, tissue samples from liver tumors identified by their basophilic appearance were punched out with sharpened cannulas and used for PCR and RFLP analysis as described above for genomic DNA. Data obtained were then compared to the p-ERK staining patterns of tumor intersections. Protein Analysis and Statistical Evaluation of the Data. Preparation of protein lysates, two-dimensional electrophoretic separation of proteins, statistical analysis and identification of significantly altered proteins spots were performed essentially as described earlier.23 In brief, frozen tissue samples were pounded in a liquid nitrogen-cooled mortar. The resulting tissue powder was resorbed in lysis buffer containing protease inhibitors, sonificated and centrifuged to finally obtain protein containing supernatant. Protein content was quantified by a standard method.24 Separation in the first dimension was carried out using IPGstrips (3-10 NL; GE Healthcare Bio-Sciences) on an IPGphor IEF-separation unit (GE Healthcare Bio-Sciences). A total of 150 µg of sample was loaded in 3 technical replicates by active ingel rehydration under 30 V for 15 h and focused at
