Volume 151, Issue 2, Pages 324-337.e12 (August 2016) Two-Step Forward Genetic Screen in Mice Identifies Ral GTPase-Activating Proteins as Suppressors of Hepatocellular Carcinoma  Takahiro Kodama, Emilie A. Bard-Chapeau, Justin Y. Newberg, Michiko Kodama, Roberto Rangel, Kosuke Yoshihara, Jerrold M. Ward, Nancy A. Jenkins, Neal G. Copeland  Gastroenterology  Volume 151, Issue 2, Pages 324-337.e12 (August 2016) DOI: 10.1053/j.gastro.2016.04.040 Copyright © 2016 AGA Institute Terms and Conditions

Figure 1 A whole-genome comprehensive catalogue of HCC CCGs. (A) Photograph showing the liver tumors found in a Liver-Onc3/HBsAg mouse. (B) Kaplan–Meier survival curves for male (left) and female (right) Liver-Onc3/HBsAg (SB/HBV), Liver-Onc3 (SB), HBsAg transgene (HBV), and control without active SB or HBV (Control) mice. The median survival for male Liver-Onc3/HBsAg mice was 56.4 weeks vs 85.4 weeks for Liver-Onc3 mice, and 100.0 weeks for HBsAg mice (log-rank test, P < 10−4). The median survival time for female Liver-Onc3/HBsAg mice was 92.0 weeks vs 100.9 weeks for HBsAg mice (log-rank test, P < 10−4). (C) Histopathologic analysis of liver tumors showed that the frequency of HCC was higher in Liver-Onc3 and Liver-Onc3/HBsAg mice compared with HBsAg mice and control mice. The overall penetrance of liver tumors (both hepatocellular adenoma [HCA] and HCC) was 100% in liver-Onc3/HBsAg and liver-Onc3 mice. (D) Trunk driver genes commonly identified in Liver-SB/HBsAg (Onc2 + Onc3) and Liver-SB mice (Onc2 + Onc3) mice. Transposon insertion sites represented by the highest sequencing read proportions (>4.5%) were selected and 62 trunk drivers common to both data sets were identified. The top 20 trunk driver genes are shown. (E) Insertion maps showing the location of transposon insertions in Rian (top), Hras (middle), or Kras (bottom) in Liver-SB/HBsAg (Onc2 + Onc3) mice and Liver-SB (Onc2 + Onc3) mice. Insertions represented by the highest read proportions (>4.5%) are shown. Each arrowhead indicates a single transposon insertion event. Red arrowheads and blue arrowheads indicate transposons inserted in the sense or antisense orientation, respectively. (F) Diagram showing our 2-step forward genetic screen approach. Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Figure 2 Acaa2, Hbs1l, and Ubr2 are TSGs in mouse and human liver. (A) SNU-398 (left) or Huh-7 (right) cells were lentivirally transduced with negative control (NC) shRNA or ACAA2 shRNA and subcutaneously injected into nude mice. Tumor volumes were measured at various different times after injection of 2.0 × 106 SNU-398 cells (N = 8 each; *P < .05) (left) or 2.0 × 106 Huh-7 cells (N = 4 for NC shRNA and 8 for ACAA2 shRNA; *P < .05) (right). (B) SNU-398 (left) or PLC/PRF/5 (right) cells were lentivirally transduced with NC shRNA or HBS1L shRNA and subcutaneously injected into nude mice. Tumor volumes were measured at various different times after injection of 2.0 × 106 SNU-398 cells (N = 16 for NC shRNA and 19 for HBS1L shRNA; *P < .05) (left) or 2.0 × 106 PLC/PRF/5 cells (N = 8 each; *P < .05) (right). (C) SNU-398 (left) or Huh-7 (right) cells were lentivirally transduced with NC shRNA or UBR2 shRNA. Tumor volumes were measured at various different times after injection of 2.0 × 106 SNU-398 cells (N = 19 for NC shRNA and 9 for UBR2 shRNA; *P < .05) (left) or 2.0 × 106 Huh-7 cells (N = 20 each; *P < .05) (right). (D–F) mRNA abundance of (D) ACAA2, (E) HBS1L, and (F) UBR2 in tumor or adjacent nontumor (NT) tissues from 372 human HCC samples including patients with the following etiologic factors (78 HBV, 32 HCV, 69 alcohol, and 93 none) registered in the TCGA database portal (*P < 0.05 vs all). Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Figure 3 RalGAPs are potent liver tumor suppressors through their suppression of the oncogenic Ral pathway. (A) SNU-398 (left), Huh-7 (center), or Hep3B (right) cells were lentivirally transduced with negative control (NC) shRNA or RALGAPA2 shRNA and the transduced cells were injected subcutaneously into nude mice. Tumor volumes were measured at various different times after injection of 1.5 × 106 SNU-398 cells (N = 8 each; *P < .05) (left), 2.0 × 106 Huh-7 cells (N = 8 each; *P < .05) (center), or 2.5 × 106 Hep3B cells (N = 10 each; *P < .05) (right). (B) Heat map showing the existence of transposon insertions in Ralgapa1, Ralgapa2, and Ralgapb in 480 tumors generated in Liver-SB/HBV (Onc2 + Onc3) and Liver-SB (Onc2 + Onc3) mice. (C) SNU-398 cells were transfected with NC siRNA or RALGAPA1 siRNA (left) or RALGAPB siRNA (right). Cell proliferation was assessed by WST-1 assay (N = 4 for RALGAPA1 siRNA and RALGAPB siRNA and N = 8 for NC siRNA; *P < .05 vs NC siRNA). (D) SNU-398 cells were lentivirally transduced with NC shRNA or RALGAPA1 shRNA (top) or RALGAPB shRNA (bottom) and the transduced cells were injected subcutaneously into nude mice. Tumor volumes were measured at various different times after injection of 2.0 × 106 SNU-398 cells containing RALGAPA1 shRNA (N = 8 each; *P < .05) (top) or RALGAPB shRNA (N = 9 each; *P < .05) (bottom). (E) One day after transfection of NC siRNA or RALGAPA2 siRNA, SNU-398 cells were treated with vehicle or 10 μmol/L RBC8 for 3 more days. Cell proliferation then was assessed by WST-1 assay (N = 4 each; *P < .05 vs all). (F) SNU-398 cells were lentivirally transduced with NC shRNA or RALGAPA2 shRNA and/or RALA and RALB shRNAs. Tumor volumes were measured at various different times after injection of 2.0 × 106 SNU-398 cells (N = 7 each; *P < .05 vs all 3 cohorts). Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Figure 4 Dual inhibition of the Ral and Raf pathways synergistically suppresses HCC cell proliferation in vitro and completely inhibits tumor growth in vivo. (A) Diagram of Ras signaling pathways and their inhibitors used in this study. (B) WST-1 assay of SNU-398 (left), Hep3B (center), and PLC/PRF/5 (right) cells after treatment with sorafenib (2 μmol/L for SNU-398 and Hep3B and 4 μmol/L for PLC/PRF/5) and/or 10 μmol/L of RBC8 for 72 hours (N = 4 each; *P < .05 vs all). (C) Cell counts of SNU-398 (left), Hep3B (center), and PLC/PRF/5 (right) cells after treatment with sorafenib (2 μmol/L for SNU-398 and Hep3B and 4 μmol/L for PLC/PRF/5) and/or 10 μmol/L of RBC8 at the indicated time courses (N = 3 each; *P < .05 vs all). (D) Clonogenic assay of SNU-398 (top), Hep3B (middle), and PLC/PRF/5 (bottom) cells after treatment with sorafenib (2 μmol/L for SNU-398 and Hep3B and 4 μmol/L for PLC/PRF/5) and/or 10 μmol/L of RBC8 for 72 hours. (E) WST-1 assay of SNU-387 (left), SNU-449 (center), and SNU-475 (right) cells after treatment with 4 μmol/L of sorafenib and/or 10 μmol/L of RBC8 for 72 hours (N = 4 each; *P < .05 vs all). (F) SNU-398 cells (2.0 × 106) were injected subcutaneously into nude mice. Once xenograft tumor volumes reached 100 mm3 (6 days after injection), mice were assigned randomly into 4 groups: control, RBC8, sorafenib, or RBC8+sorafenib. Mice then were treated with oral gavage of 50 mg/kg RBC8 and/or intraperitoneal injection of 50 mg/kg sorafenib 5 days a week for 12 days. Xenograft tumor volumes then were measured at different times after the initiation of treatment (N = 4 for RBC8+sorafenib and 3 for all other cohorts; *P < .05 vs all). Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Supplementary Figure 1 (A–I) Kaplan–Meier plot of patient survival based on mRNA abundance in tumor tissue of 221 human HCC samples (GSE14520). Statistical analysis was performed using the log-rank test. (A) ACCA2, (B) ALDH2, (C) DMXL1, (D) HBS1L, (E) KIF1B, (F) STK11, (G) UBR2, (H) ZBTB20, and (I) ZCCHC6. Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Supplementary Figure 2 (A–H) WST-1 assay results of individual siRNA oligos targeting Acaa2 in (A) SNU-398 cells and (B) Huh-7 cells, (C) Hbs1l in SNU-398 cells, and (D) PLC/PRF/5 cells, Ralgapa2 in (E) SNU-398 cells and (F) Huh-7 cells, and Ubr2 in (G) SNU-398 cells and (H) Huh-7 cells (*P < .05 vs negative control [NC] siRNA; N = 4). Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Supplementary Figure 3 (A–D) Mouse progenitor IFLCs were transduced lentivirally with negative control (NC) shRNA (A) or Acaa2 shRNA, (B) Hbs1l shRNA, (C) Ralgapa2 shRNA, or (D) Ubr2 shRNA. Transduced cells then were injected subcutaneously into nude mice and the mice were monitored for tumor formation. Tumor volumes were measured at different times after subcutaneous injection of 1.0 × 106 IFLCs (A) (N = 4) or (B–D) (N = 10). (E and F) mRNA expression levels of ACAA2 were determined by quantitative PCR after transduction with NC shRNA or ACAA2 shRNA in (E) SNU-398 or (F) Huh-7 cells (N = 4; *P < .05). (G) Protein expression levels of ACAA2 were determined by Western blot (WB) after transduction with NC shRNA or ACAA2 shRNA in SNU-398 or Huh-7 cells. (H and I) mRNA expression levels of HBS1L were determined by quantitative PCR after transduction with NC shRNA or HBS1L shRNA in (H) SNU-398 or (I) PLC/PRF/5 cells (N = 4; *P < .05). (J) Protein expression levels of HBS1L were determined by WB after transduction with NC shRNA or HBS1L shRNA in SNU-398 or PLC/PRF/5 cells. (K and L) mRNA expression levels of UBR2 were determined by quantitative PCR after transduction with NC shRNA or UBR2 shRNA in (K) SNU-398 or (L) Huh-7 cells (N = 4; *P < .05). (M) Protein expression levels of UBR2 were determined by WB after transduction with NC shRNA or UBR2 shRNA in SNU-398 or Huh-7 cells. Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Supplementary Figure 4 (A–F) The relative expression level of RALGAPA2 was measured by quantitative PCR in (A and D) SNU-398, (B and E) Huh-7, or (C and F) Hep3B cells (N = 4; *P < .05) after lentiviral transduction with negative control (NC) shRNA or RALGAPA2 shRNAs. (G) Protein expression levels of RALGAPA2 were determined by Western blot (WB) after transduction with NC shRNA or RALGAPA2 shRNA in SNU-398, Huh-7, or Hep3B cells. (H) SNU-398, (I) Huh-7, or (J) Hep3B cells were transduced lentivirally with NC shRNA or RALGAPA2 shRNA. Transduced cells were injected subcutaneously into nude mice and tumor growth was measured at different times after subcutaneous injection of (H) 1.5 × 106 SNU-398 cells (N = 8 for NC shRNA and N = 10 for RALGAPA2 shRNA; *P < .05); (I) 2.0 × 106 Huh-7 cells (N = 8; *P < .05); or (J) 2.5 × 106 Hep3B cells (N = 10; *P < .05). Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Supplementary Figure 5 (A) mRNA expression level of RALGAPA1 was determined by quantitative PCR after lentiviral transduction of SNU-398 cells with negative control (NC) shRNA or RALGAPA1 shRNA (N = 4; *P < .05). (B) Protein expression levels of RALGAPA1 were determined by Western blot (WB) after transduction with NC shRNA or RALGAPA1 shRNA in SNU-398 cells. (C) mRNA expression level of RALGAPB was determined by quantitative PCR after lentiviral transduction of SNU-398 cells with NC shRNA or RALGAPB shRNA (N = 4; *P < .05). (D) Protein expression levels of RALGAPB were determined by WB after transduction with NC shRNA or RALGAPB shRNA in SNU-398 cells. (E) mRNA expression level of RALGAPA2, RALA, or RALB in SNU-398 cells was determined by quantitative PCR after lentiviral transduction with shRNAs targeting each of these genes (N = 4; *P < .05 vs NC shRNA). (F) Protein expression level of RALGAPA2, RALA, or RALB in SNU-398 cells was determined by WB after lentiviral transduction with shRNAs targeting each of these genes. Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Supplementary Figure 6 Protein expression level of Ras downstream pathways was determined by Western blot (WB) after treatment with 2 μmol/L sorafenib and/or 10 μmol/L of RBC8 for 24 hours. Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Supplementary Figure 7 WST-1 assay of (A) SNU-398 and (B) Hep3B cells after treatment with 2 μmol/L sorafenib and different dosages of RBC8 for 72 hours (N = 3 or N = 4 each; *P < .05 vs all). Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Supplementary Figure 8 (A) Gene ontology analysis of 1917 high-confidence CCGs determined for PANTHER GO-Slim Biological Processes. P values for enrichment are represented on a -log scale. The numbers next to each bar indicate the number of CCGs found associated with each biological process. (B) Gene ontology analysis of 27 TSGs validated by shRNA library screening determined for PANTHER GO-Slim Biological Processes. Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions

Supplementary Figure 9 (A–C) mRNA abundance of (A) RALGAPA2, (B) RALGAPA1, or (C) RALGAPB in tumor or adjacent nontumor (NT) tissues from 372 human HCC samples, including patients with the following etiologic factors: 78 HBV, 32 HCV, 69 alcohol, and 93 none, registered in the TCGA database portal (*P < .05 vs all). Gastroenterology 2016 151, 324-337.e12DOI: (10.1053/j.gastro.2016.04.040) Copyright © 2016 AGA Institute Terms and Conditions