Volume 143, Issue 3, Pages 787-798.e13 (September 2012) Hepatoma Cells From Mice Deficient in Glycine N-Methyltransferase Have Increased RAS Signaling and Activation of Liver Kinase B1  Nuria Martínez–López, Juan L. García–Rodríguez, Marta Varela–Rey, Virginia Gutiérrez, David Fernández–Ramos, Naiara Beraza, Ana M. Aransay, Karin Schlangen, Juan Jose Lozano, Patricia Aspichueta, Zigmund Luka, Conrad Wagner, Matthias Evert, Diego F. Calvisi, Shelly C. Lu, José M. Mato, María L. Martínez–Chantar  Gastroenterology  Volume 143, Issue 3, Pages 787-798.e13 (September 2012) DOI: 10.1053/j.gastro.2012.05.050 Copyright © 2012 AGA Institute Terms and Conditions

Figure 1 Characterization of OKER cells. Graphical representation (mean ± SEM) of (A) mRNA expression (arbitrary units) of the indicated genes. *P < .05, OKER and GNMT-KO hepatocytes versus WT mouse hepatocytes. (B) Intracellular levels of SAMe. *P < .05, OKER and GNMT-KO hepatocytes versus WT hepatocytes. (C and D) mRNA expression (arbitrary units) of the indicated genes. *P < .05, OKER and GNMT-KO hepatocytes versus GNMT-WT mouse hepatocytes. (E) Total protein extracts from OKER, GNMT-KO, and GNMT-WT mouse hepatocytes were analyzed via Western blotting with the indicated antibodies. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Figure 2 Ras/MEK/ERK-mediated p-LKB1 (Ser428) regulation. (A) OKER cells were cultured with (A) sorafenib (10 μmol/L), (B) U0126 (10 μmol/L), (D) BI-D1870 (20 μmol/L), and (E) H89 (25 μmol/L). Whole cell lysates were analyzed via Western blotting. The extracts were immunoprecipitated with (C) AMPKα1 or (F) p-LKB1 (Ser428) antibodies. Immunoprecipitates (IP) and lysates (INPUT) were analyzed via WB. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Figure 3 5′-Azacytidine inactivates the Ras pathway in OKER cells. OKER cells and GNMT-WT and GNMT-KO hepatocytes were treated with 5′-azacytidine (10 μmol/L). (A) Graphical representation of the mRNA expression. *P < .05 treated versus untreated cells. (B) Ras activity was assessed and probed with anti-RAS antibody. (C and D) Cytosolic and whole cell extracts of OKER cells and GNMT-WT and GNMT-KO hepatocytes were analyzed via Western blotting. (E) OKER cells transfected with control or Bcl-2 siRNA and treated with 5′-azacytidine for 12 hours were analyzed by Western blotting. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Figure 4 Influence of LKB1 and CaMKKβ-mediated AMPK response to 5′-azacytidine in the OKER cells. OKER cells were transfected with control (A) AMPKα1 or (B) LKB1 siRNA prior treatment with 10 μmol/L 5′-azacytidine. Cell lysates were analyzed via Western blotting. OKER cells were incubated with (C) STO-609 (10 μmol/L) or (F) forskolin (10 μmol/L) for 1 hour before 5′-azacytidine treatment for the indicated times. OKER cells were transfected with (D) pcDNA3-Flag-CaMKKα or pcDNA3-Flag-CaMKKβ or (E) control, CAMKKα, or CaMKKβ siRNAs and treated with 5′-azacytidine. Whole cell lysates were analyzed via Western blotting. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Figure 5 5′-Azacytidine attenuates tumor growth. A total of 2 × 106 OKER-GFP cells were injected subcutaneously as described in Materials and Methods. (A) Graphical representation of tumor volume. *P < .05, 5′-azacytidine versus control. Paraffin-embedded tumor sections were stained with (B) H&E, CD31 counterstained with hematoxylin, and terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling (TUNEL) assay, and (E) p-LKB1 (Ser428), p-AMPKα (Thr172), and p-p53 (Ser15) (left). Graphical representation of the quantitative analysis for each staining. *P < .05, 5′-azacytidine versus control (right). Original magnification, 200×. (C) Ras activity was assessed and probed with anti-RAS antibody. (D) Fold change graphical representation in the Ras activity. *P < .05, 5′-azacytidine versus control. (F) Whole tissue extracts from tumors were analyzed by Western blotting. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Figure 6 LKB1 ablation attenuates tumor growth. Graphical representation of (A) changes in tumor size, (B) tumor size increment, and (C) tumor weight at the end of the experiment. *P < .05 and #P < .05 (siRNA LKB1 versus siRNA control and siRNA LKB1 + 5′-azacytidine, respectively). (D) Paraffin-embedded sections of tumors were stained with corresponding antibodies (left). Graphical representation of the quantitative analysis for each staining. *P < .05 versus control siRNA (right). (E) Ras activity was assessed as described before. (F) Fold change graphical representation in Ras activity (right). *P < .05 siRNA LKB1 versus siRNA control. #P < .05, siRNA LKB1 versus siRNA LKB1 + 5′-azacytidine. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Figure 7 GNMT, LKB1, and Ras activity correlation. (A) Box plots of GNMT (fold change, −3.04; P [Student] < .0001) and RASGRP3 (fold change, +1.29; P [Student] < .0001) levels in HCC (225 samples) and nontumoral human samples (200). (B) Pearson correlation with an r of −0.308 and associated P value of <1 × 10−11 between GNMT and RASGRP3 genes. (C) Representation of GNMT and STK11 expression levels in 29 HCC with better prognosis (HCCb) and 27 HCC with poor prognosis (HCCp). (D) Pearson correlation with an r of −0.71 and associated P value of <1 × 10−16 between GNMT and STK11 in the normal liver, surrounding liver near the tumor, HCCb, and HCCp. (E) Protein expression was analyzed via Western blotting from the human samples at different states. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Supplementary Figure 1 Characterization of OKER cells. (A) Signaling pathways regulated in the OKER cell line compared with primary WT mouse hepatocytes. (B) Graphical representation of the densitometric analysis of the proteins presented in Figure 1E. *P < .05, OKER and GNMT-KO hepatocytes versus WT mouse hepatocytes. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Supplementary Figure 2 Role of DNMT in OKER cells and GNMT WT and KO hepatocytes. Graphical representation (mean ± SEM) of (A) messenger RNA expression (arbitrary units) of the indicated genes. *P < .05, OKER cells and GNMT KO hepatocytes versus WT mouse hepatocytes. DNMT activity measurements of the OKER cells and GNMT-KO hepatocytes versus WT mouse hepatocytes after (B) 12 hours and (C) 24 hours of 5-azacytidine treatment. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Supplementary Figure 3 Graphical representation (mean ± SEM) of the densitometric analysis of PARP cleavage observed in (A) Figure 4A (left) and (B) Figure 4B (right). *P < .05, AMPKα1 or LKB1 siRNA versus control siRNA transfected cells. (B) OKER cells were transfected with WT-LKB1 or kinase-dead LKB1 (K78I) and whole cell extracts were analyzed via Western blotting for the indicated antibodies. (C) Western blot analysis of HepG2, Huh7, and PLC cell lines for the indicated antibodies in whole cell extract. (D) Caspase-3 activity and Western blot analysis in whole extract of HepG2, Huh7, and PLC cell lines after silencing LKB1 with siRNAs. (E) Western blot analysis in whole extract of HepG2, Huh7, and PLC cell lines with the detailed antibodies after LKB1 overexpression with a pcDNA3.1 vector. (F) Caspase-3 activity and Western blot analysis in whole extract of HepG2, Huh7, and PLC cell lines after GNMT overexpression with pCHE expression vector. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Supplementary Figure 4 (A) Graphical representation (mean ± SEM) of the densitometric analysis of p-AMPKα (Thr172)/GAPDH represented in Figure 4D and E. *P < .05 versus pcDNA3 untreated cells. #P < .05 versus pcDNA3 cells treated with 5′-azacytidine. (B) Graphical representation (mean ± SEM) of the messenger RNA expression of the indicated genes in Figure 4E; *P < .05 versus siRNA control. (C) Graphical representation (mean ± SEM) of the densitometric analysis of PARP/β-actin (left) and p-AMPKα (Thr172)/β-actin (right) represented in Figure 4F. *P < .05 versus untreated cells. #P < .05, 5′-azacytidine versus forskolin + 5′-azacytidine-treated cells. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Supplementary Figure 5 Regulation of the Ras activity after silencing LKB1. (A) Ras activity was assessed by immunoprecipitation with anti-Raf1 RBD agarose and probed with anti-RAS antibody from OKER cell total protein extracts. (B) mRNA expression (arbitrary units) of RASGRP3 gene; *P < .05 control versus LKB1 siRNA transfected OKER cells. (C) Fold changes of RASGRP3 expression levels (arbitrary units); *P < .05, control versus RASGRP3 siRNA transfected OKER cells. (D) Dot plot showing incorporation of PI (y-axis) versus Annexin V (x-axis) evidences higher cell death in RASGRP3 versus control siRNA transfected cells. (D) Graphical representation (mean ± SEM) of the dot plot analysis differentiating between early and late apoptosis. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Supplementary Figure 6 (A) Graphical representation of the densitometric analysis for proteins represented in Figure 7B from human HCC with better prognosis; *P < .05 versus human HCC with poorer prognosis. (B) Logistic regression performed to quantify the predictability of a GNMT and STK11 model in better and poorer prognosis of HCC. Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions

Supplementary Figure 7 Schematic representation of proliferative pathways in WT hepatocytes versus GNMT-deficient HCC. (A) The Ras/MEK/ERK and the noncanonical LKB1/AMPK/HuR pathway drive WT hepatocyte proliferation in parallel. (B) In GNMT-deficient HCC, however, the epigenetic effect of SAMe in the RASSF1 gene promotes WT Ras oncogene hyperactivity. The Ras signaling cascade and cAMP/PKA pathway are responsible for LKB1 hyperphosphorylation (Ser428), uncoupling LKB1 and AMPK. LKB1 plays an antiapoptotic role by regulating the Ras oncogene activity in a positive feedback loop that involves the up-regulation of the RASGRP3 messenger RNA expression. The treatment with the demethylating agent 5′-azacytidine induces apoptosis by (1) recovering the expression of the RASSF1 gene, which (2) blocks Ras activity, (3) down-regulating the Ras signaling pathway leading to dephosphorylation of LKB1 (Ser428) with the subsequent blockade of Ras activity and (4) inhibiting the cAMP/PKA cascade leading to CamKKβ-mediated activation of apoptotic AMPKα (Thr172). Gastroenterology 2012 143, 787-798.e13DOI: (10.1053/j.gastro.2012.05.050) Copyright © 2012 AGA Institute Terms and Conditions