Supplementary Materials and Methods

Animals and treatments.

F344 and BN rats (Charles-River-Italia, Calco, Italy) were housed individually in a room with constant temperature (22°C) and humidity (55%) and with a 12 h light (6 a.m-6 p.m.)-dark cycle, and treated according to the ‘‘resistant hepatocyte’’ protocol [1] consisting of a 150 mg/kg intraperitoneal dose of diethylnitrosamine followed by 15 days of feeding a 0.02% 2-acetylaminofluorene-containing hyperprotein diet, with a partial hepatectomy at the midpoint of this feeding regime. Rats were killed by bleeding through thoracic aorta, under ether anaesthesia. Freshly removed livers were serially sectioned with 〜0.5 cm intervals. Thirty-two weeks after initiation, dysplastic nodule (DN) macroscopically identified by their sharp grayish-white color, were scooped out from the liver, free of surrounding parenchyma (as shown by histological control). At 45-50 and 60-67 weeks, HCC were collected from F344 and BN rats, respectively, leaving out a small rim of neoplastic tissue. Only DNs with diameter ≧3 mm were collected from both rat strains andsplit in half. One half of this material was processed for histology, histochemistry, and immunohistochemistry, and the other half was stored at -80°C. Histological (HE staining), histochemical (silver staining of reticulin) and immunohistochemical (glutamine synthase immunostaining) criteria were used, in addition to morphology, to classify liver lesions according to the published criteria [2-4]. The average diameters of lesions were: 3.8 ± 0.8, 5.8 ± 1.6, and 14.4 ± 6.7mm for low-grade and high-grade dysplastic nodules and HCC, respectively. Animals received human care, and study protocols were in compliance with the National Institutes of Health guidelines for use of laboratory animals.

Human Tissue Samples

Five normal livers and60 HCC were used. Supplementary Table S1shows patients’ clinicopathologic features. Liver tissues were archival samples provided by the Departments of Surgery of the University “La Sapienza” of Roma, and of the University of Sassari, Italy. Informed patients’ consent and Institutional Review Board approval was obtained at these Departments.

Histology and immunohistochemistry.

DN and HCC were processed for glutamine synthase as published [5]. Labeling index (LI) was determined after i.p. injection into the rats of 5 mg/100 g body wt of 2-bromo-3-deoxyuridine (BrdU), 2 h before killing, and the percentage of nuclear BrdU incorporation was evaluated by the ‘cell proliferation kit’ (Amersham Biosciences, Cologno Monzese, Italy). Apoptotic index (AI) was the percentage ofapoptotic bodies stained by hematoxylin/eosin, or nuclear changes representing apoptosis (chromatin condensation, margination, and fragmentation) revealed by propidium iodide. LI and AI in rat liver tissues were evaluated by scoring 3000 normal hepatocytes or preneoplastic or neoplastic cells per liver. The staining results were independently and blindly evaluated by two senior pathologists (GM and FF).

Microarray analysis.

Four frozen rat livers, 8 pools of 5-10 DNs/rat, and 7 single HCC, from each strain, were dissected into 〜9 µm thick sections and immediately lysed in RNA extraction buffer: Total RNA isolation and DNase digestion were performed by the RNA Easy Mini kit (Qiagen S.p.A., Milano, Italy). RNA integrity was determined by gel electrophoresis. Total RNA from control livers, DNs, and HCCs was reversed, fluorescently labeled with Cy3 or Cy5 by Low RNA Unput Linear AMP Kit (Agilent Technologies, Wilmington, DE), and hybridized using Agilent In situ Hybridization Kit-plus (Agilent Technologies). Fluorescence intensities were measured by a dual-laser ScanArray Express Microarray Scanner (Perkin Elmer Life Sciences, Boston, MA). Spot subsets were selected by the criterion of minimum variance in duplicate fluorescence ratio measurements, when the fluorescence signal was higher than 0.3% of the measurable total signal dynamics range above background in both channels of the hybridization. The intensity of each spot was normalized to average intensity of housekeeping genes. Average expression ratios were transformed to log2. Genes with expression ratio at least 2 fold different from that of reference samples were selected. The algorithm based on Pearson correlation coefficients was used for hierarchical cluster analysis. K-mean clustering analysis, and visualization of analyzed data were performed as described [6].

Quantitative Real-Time Reverse-Transcription Polymerase Chain Reaction.

Primers for quantiative Real-time RT-PCR (qPCR) were chosen by the “Assays-on-DemandTM Products” (Applied Biosystems, Foster City, CA).Polymerase chain reaction was performed with 75-300 ng of complementary DNA using an ABI Prism 7000 and Taq Man Universal PCR Master Mix (Applied Biosystems). Cycling conditions were: 10 minutes of denaturation at 95°C and 40 cycles at 95°C for 15 seconds and at 60°C for 1 minute. Quantitative values were calculated using PE Biosystems Analysis software (PE Biosystems, San Jose, CA)

Western Blot Analysis.

Hepatic tissue samples and cell suspensions from cultured mouse cancer cells were homogenized in lysisbuffer [30 mM Tris (pH 7.5), 150 mM NaCl, 1% NP-40, 0.5% Nadeoxycholate, 0.1% SDS, 10% glycerol, and 2 mM EDTA] containingthe Complete Protease Inhibitor Cocktail (Roche Molecular Biochemicals, Indianapolis, IN)and sonicated. Protein concentrations were determined with theBio-Rad Protein Assay Kit (Bio-Rad, Hercules, CA) using bovine serum albuminas standard. Membranes wereprobed with mouse monoclonal antibodies against c-Myc and Bhmt (Santa Cruz Biotechnology: SC-42,SC-69708; Santa Cruz, CA), rabbit polyclonal antibodies against Gnmt and Dusp1 (Mkp-1),(Santa Cruz Biotechnology: SC-68871, SC-1199),PP2A (Cell Signalling 2038), and β-Actin(Abcam AB-636276; Abcam Cambridge, UK). Membranes were thenprocessed as reported [5]. Each primary antibody was followed by incubation with horseradishperoxidase-secondary antibody diluted 1:5000 for 1 h and thenrevealed with the Super Signal West Pico Chemiluminescence Substrate Kit (Pierce Chemical Co.,New York, NY). For each protein, densities were calculated by ImageQuaNT 5.1 software (GE Healthcare, Piscataway, NJ), normalized to β-Actin (Santa Cruz Biotechnology, Santa Cruz, CA; dilution 1:10000) levels and mean values evaluated for statistical significance.

Statistical analysis.

Data are expressed as means ± SD. GraphPad Prism 5.01 ( was used to evaluate, by Tukey-Kramer test, the significance of differences between means of qPCR and Western blot analyses of rat samples, and calculate the correlation coefficient (R) by multiple regression analysis. Assumption that the data are sampled from populations that follow Gaussian distribution was evaluatedusing the D’Agostino and Pearsons normality test. Differences in gene expression of human HCC subgroups were evaluated by the Mann-Whitney U-test. The probability of overallsurvival was estimated accordingto Kaplan-Meier and Log-rank (Mantel-Cox) test. Cox proportional hazards regression method was used to estimate the predictivity of patients’ survival.Statistical analysis of microarray results by parametric Student’s t-test and the False Discovery Rate (FDR) method to correct p-values and control false identifications, were performed by BRB Array Tools ( Values of p < 0.05 were considered significant.

References

  1. Solt DB, Medline A, and Farber E. Rapid emergence of carcinogen-induced hyperplastic lesions in a new model for the sequential analysis of liver carcinogenesis. Am J Pathol 88, 595-618(1977).
  2. Squire RA, Levitt MH. Report of a workshop on classification of specific hepatocellular lesions in rats. Cancer Res 35, 3214-3223 (1975).
  3. Terminology of nodular hepatocellular lesions. International Working Party. Hepatology22, 983-993 (1995).
  4. Sakamoto M. Early HCC: diagnosis and molecular markers. J Gastroenterol 44[Suppl XIX], 108–111 (2009)
  5. Pascale RM, Simile MM, Calvisi DF, Frau M, Muroni MR, Seddaiu MA, Daino L, Muntoni MD, De Miglio MR, Thorgeirsson SS, and Feo F.Role of the modulation of p16INK4A and E2f4 activity by Cdc37, Hsp90 and Crm1 in the phenotypic behavior of preneoplastic and neoplastic liver lesions of rat strains with different genetic predisposition to hepatocarcinogenesis and in human liver cancer. Hepatology42, 1310-1319 (2005).
  6. Bittner M, Meltzer P, Chen Y, Jiang Y, Seftor E, Hendrix M, Radmacher M, Simon R, Yakhini Z, Ben-Dor A, Sampas N, Dougherty E, Wang E, Marincola F, Gooden C, Lueders J, Glatfelter A, Pollock P, Carpten J, Gillanders E, Leja D, Dietrich K, Beaudry C, Berens M, Alberts D, and Sondak VMolecular classification of cutaneous malignant melanoma by gene expression profiling. Nature 403,503-511 (2000).