Scholarly article on topic 'Dual Shp2 and Pten Deficiencies Promote Non-alcoholic Steatohepatitis and Genesis of Liver Tumor-Initiating Cells'

Dual Shp2 and Pten Deficiencies Promote Non-alcoholic Steatohepatitis and Genesis of Liver Tumor-Initiating Cells Academic research paper on "Biological sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Cell Reports
OECD Field of science
Keywords
{"liver tumorigenesis" / "non-alcoholic steatohepatitis" / "tumor-initiating cells" / "tumor suppressor" / PTEN / Shp2}

Abstract of research paper on Biological sciences, author of scientific article — Xiaolin Luo, Rui Liao, Kaisa L. Hanley, Helen He Zhu, Kirsten N. Malo, et al.

Summary The complexity of liver tumorigenesis is underscored by the recently observed anti-oncogenic effects of oncoproteins, although the mechanisms are unclear. Shp2/Ptpn11 is a proto-oncogene in hematopoietic cells and antagonizes the effect of tumor suppressor Pten in leukemogenesis. In contrast, we show here cooperative functions of Shp2 and Pten in suppressing hepatocarcinogenesis. Ablating both Shp2 and Pten in hepatocytes induced early-onset non-alcoholic steatohepatitis (NASH) and promoted genesis of liver tumor-initiating cells likely due to augmented cJun expression/activation and elevated ROS and inflammation in the hepatic microenvironment. Inhibiting cJun partially suppressed NASH-driven liver tumorigenesis without improving NASH. SHP2 and PTEN deficiencies were detected in liver cancer patients with poor prognosis. These data depict a mechanism of hepato-oncogenesis and suggest a potential therapeutic strategy.

Academic research paper on topic "Dual Shp2 and Pten Deficiencies Promote Non-alcoholic Steatohepatitis and Genesis of Liver Tumor-Initiating Cells"

Cell Reports

Article

Dual Shp2 and Pten Deficiencies Promote Nonalcoholic Steatohepatitis and Genesis of Liver Tumor-Initiating Cells

Graphical Abstract

Authors

Xiaolin Luo, Rui Liao, Kaisa L. Hanley.....

Rohit Loomba, Shuang-Jian Qiu, Gen-Sheng Feng

Correspondence

gfeng@ucsd.edu

In Brief

Shp2 antagonizes Pten in leukemogenesis. Luo et al. now find that Shp2 and Pten synergistically suppress carcinogenesis in the liver. Ablating Shp2 and Pten in hepatocytes induced early-onset non-alcoholic steatohepatitis (NASH) and promoted genesis of tumor-initiating cells. These findings suggest a mechanism underlying disease as well as a therapeutic strategy.

Highlights Accession Numbers

• Shp2/Ptpn11 cooperates with Pten to maintain liver GSE74732

homeostasis and functions

• Dual deletion of Shp2 and Pten induces NASH-driven hepatocarcinogenesis

• Combined Shp2/Pten deficiencies promote genesis of liver tumor-initiating cells

• Inhibiting c-Jun suppressed liver cancer without improving NASH

Luo et al., 2016, Cell Reports 17, 2979-2993 ciossMark December 13, 2016 © 2016 The Authors.

http://dx.d0i.0rg/l 0.1016/j.celrep.2016.11.048

CelPress

Cell Reports

Article

Dual Shp2 and Pten Deficiencies Promote Non-alcoholic Steatohepatitis and Genesis of Liver Tumor-Initiating Cells

Xiaolin Luo,1 Rui Liao,135 Kaisa L. Hanley,2 Helen He Zhu,1 Kirsten N. Malo,2 Carolyn Hernandez,4 Xufu Wei,15 Nissi M. Varki,1 Nazilla Alderson,1 Catherine Chu,2 Shuangwei Li,1 Jia Fan,3 Rohit Loomba,4 Shuang-Jian Qiu,3 and Gen-Sheng Feng12 6 *

department of Pathology, School of Medicine 2Molecular Biology Section, Division of Biological Sciences University of California, San Diego, La Jolla, CA 92093, USA

3Liver Cancer Institute, Zhongshan Hospital, Fudan University, Shanghai 200032, China

4Department of Medicine, NAFLD Research Center, University of California, San Diego, La Jolla, CA 92093, USA 5Department of Hepatobiliary Surgery, The First Affiliated Hospital, Chongqing Medical University, Chongqing 40016, China 6Lead Contact

'Correspondence: gfeng@ucsd.edu http://dx.doi.org/10.1016/j.celrep.2016.11.048

SUMMARY

The complexity of liver tumorigenesis is underscored by the recently observed anti-oncogenic effects of oncoproteins, although the mechanisms are unclear. Shp2/Ptpn11 is a proto-oncogene in hematopoietic cells and antagonizes the effect of tumor suppressor Pten in leukemogenesis. In contrast, we show here cooperative functions of Shp2 and Pten in suppressing hepatocarcinogenesis. Ablating both Shp2 and Pten in hepatocytes induced early-onset non-alcoholic steatohepatitis (NASH) and promoted genesis of liver tumor-initiating cells likely due to augmented cJun expression/activation and elevated ROS and inflammation in the hepatic microenvironment. Inhibiting cJun partially suppressed NASH-driven liver tumorigenesis without improving NASH. SHP2 and PTEN deficiencies were detected in liver cancer patients with poor prognosis. These data depict a mechanism of hepato-oncogenesis and suggest a potential therapeutic strategy.

INTRODUCTION

Primary liver cancer, mainly hepatocellular carcinoma (HCC), has become the second leading cause of cancer-related death worldwide (Theise, 2014) . Although its etiology is strongly associated with hepatitis B or C virus (HBV or HCV) infections in Asian and African countries, the rapidly rising incidences in the Western world are linked to obesity, alcohol abuse, and non-alcoholic fatty liver disease (NAFLD) or steatohepatitis (NASH). NASH is one major manifestation of metabolic disorders, which trigger hepatic injuries, inflammation, fibrosis, and carcinogenesis (Lade et al., 2014; Michelotti et al., 2013).

Genetic analyses of human HCC samples have detected aberrant activation of classical oncogenic signaling pathways, such as MET, Wnt/p-catenin, NF-kB, and JNK (Kaposi-Novak et al., 2006; Pilati et al., 2014; Zucman-Rossi et al., 2015). Deficient expression or inactivating mutations were also detected in classical tumor suppressor genes, such as p53, RB, p21, and p27 in human liver cancers (Zucman-Rossi et al., 2015). Pten (phosphatase and tensin homolog deleted from chromosome 10), a lipid phosphatase for phosphatidylinositol 3,4,5-trisphosphate, is a classical tumor suppressor that counteracts PI3K signaling (Cantley, 2002; Song et al., 2012; Worby and Dixon, 2014). Pten deficiency or mutations were detected in various types of human cancer, including HCC (Chalhoub and Baker, 2009), and targeted Pten deletion in hepatocytes led to spontaneous development of HCC or ICC (intrahepatic cholangiocarcinoma) in mice (Galicia et al., 2010; Horie et al., 2004; Kenerson et al., 2013).

The above clinical and experimental data argue that liver tumorigenesis shares common mechanisms, i.e., over-activation of proto-oncogenes and/or loss/inactivation of tumor suppressors resulting in neoplastic cell proliferation. Accordingly, great efforts have been devoted to the development of pharmaceutical compounds that disrupt the classical oncogenic pathways for liver cancer therapy, with Sorafenib, amultikinase inhibitor, as the most widely used drug (Llovet et al., 2008). Unfortunately, Sorafenib and other similar oncoprotein inhibitors have achieved very little therapeutic benefit for HCC patients (Llovet et al., 2008). The systematic failure in the "mechanism-based" HCC therapy is evidently due to inadequate understanding of the complexity in hepatocarcinogenesis. One interesting finding made by several groups recently is the unanticipated anti-oncogenic effect of classical oncoproteins (Feng, 2012). Deletion of Met, Egfr, ctnnbl, or Ikkb in hepatocytes surprisingly enhanced HCC development in mice treated with a chemical carcinogen diethylnitrosamine (DEN) (Feng, 2012; Lanaya et al., 2014; Maeda et al., 2005; Takami et al., 2007; Zhang et al., 2010). These data may explain why inhibiting classical

Cell Reports 17, 2979-2993, December 13,2016 © 2016 The Authors. 2979 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

oncogenic pathways achieved little therapeutic effect for HCC patients, and also unlock the previously unrecognized complexity of liver tumorigenesis, which urgently needs elucidation.

Shp2/Ptpn11 is the first identified oncogenic tyrosine phos-phatase, with dominantly activating mutations detected in several types of leukemia (Chan and Feng, 2007), and Shp2 positively regulates Ras-Erk signaling (Lai et al., 2004; Neel et al., 2003). Indeed, ablating Shp2 abrogates myeloproliferative neoplasm induced by Pten loss, indicating opposing effects between the two molecules in myeloid cells (Zhu et al., 2015). However, in contrast to its pro-leukemogenic effect, deleting Shp2 in hepatocytes triggered hepatocellular adenoma (HCA) in aged mice and also enhanced DEN-induced HCC development (Bard-Chapeau et al., 2011). Nevertheless, the concept that these oncoproteins act as tumor suppressors in the liver has not been well accepted, due to concerns on the relevance of DEN-induced animal models to HCC patients. We choose to dissect the anti-oncogenic role of Shp2 in the liver by determining its functional interaction with a classical tumor suppressor Pten.

RESULTS

Simultaneous Deletion of Shp2 and Pten Cooperatively Promotes Liver Tumorigenesis

We generated a hepatocyte-specific Pten and Shp2 doubleknockout (DKO) mouse line {PtenmL.Shp2mL.Alb-Cre+), by crossing Shp2fl/fl (Zhang et al., 2004) and Ptenfl/fl (Lesche et al., 2002) mice with Albumin-Cre transgenic mice. In this study, we performed comparative phenotypic analysis of DKO with SKO (Shp2hep-/-, or Shp2fl/flAlb-Cre+), PKO (Ptenhep-/-, or Pten2fl/fL.Alb-Cre+), and wild-type (WT) (Pten2mL.Shp2mLAlb-Cre~) mice. Deletion of either Shp2 or Pten alone did not cause liver tumors until late stages (Figures 1A and 1B). However, 80% of DKO mice spontaneously developed liver tumors in 5 months, and the tumor incidences reached 100% at 7 months of age. By evaluating tumor numbers, sizes (maximal tumor diameter), and liver-to-body weight ratios, we found that ablating both Shp2 and Pten triggered more severe liver tu-morigenesis than deleting either gene alone at all time points examined.

Pathological examination of 60 WT mice in total revealed no hepatic tumor by 18 months of age (Table S1). We detected HCA in two of six SKO mice at 12 months and in six of seven of the mice at 18 months, consistent with previous observation (Bard-Chapeau et al., 2011). By examining 12 PKO mice at 7 months, we detected ICC in one and HCA in three, with no tumor nodules observed before 7 months. Tumor incidences rose up to 100% (11 of 11) in PKO mice at the age of 12 months, and all tumors were either HCA or ICC, with no HCC observed. At 16 months of age, three of seven PKO animals developed HCC. Notably, deleting both Shp2 and Pten dramatically accelerated tumor initiation and malignant transition. We detected HCA in six of nine and HCC in two of nine DKO mice examined at 4-5 months, and six of ten DKO mice had HCC at 7 months. We also observed mixed ICC/HCC tumors in DKO mice, and three DKO mice with HCC exhibited lung metastasis at 12 months (Figure 1E).

Histological analysis showed similar pathological features for tumors from PKO and DKO livers (Figures 1C and 1D), and HCAs in both PKO and DKO showed tumor cells with large lipid droplets in the cytoplasm and the nucleus against the cell membrane. HCCs exhibited a trabecular pattern and had less lipids in cells and stained positive for HepPar-1. ICCs displayed well-differentiated tubules lined with one or multiple layers of cuboidal epithelial cells that were CK-19-positive. Tumors with mixed HCC/ICC phenotype stained partly positive for HepPar-1 and CK-19. PCR analysis of genomic DNA and immunoblotting of proteins confirmed efficient deletion of Shp2 and/or Pten in both tumor and non-tumor tissues in respective animals (Figure 1F). Immunoblotting also detected markedly elevated levels of Cyclin D1 and PCNA in both PKO and DKO tumors relative to tumor-surrounding tissues, with no significant difference observed for c-Myc and p-catenin expression among the four groups of mice.

Combined Shp2 and Pten Deficiencies in Hepatocytes Trigger Early-Onset NASH

PKO mice developed hepatomegaly and NAFLD at 2 months, consistent with previous reports (Horie et al., 2004; Stiles et al., 2004), but DKO animals had even more severely enlarged livers at the same age and their liver-to-body weight ratios increased significantly compared to the other three groups (Figures S1A and S1B). Quantification of Oil-Red-O staining and triglyceride levels showed comparable fat deposition in PKO and DKO mice; histological analysis revealed hepatocytes with fat pellets in both peri-central and peri-portal areas in DKO, whereas steatosis was detected only in the peri-central area in PKO mice (Figures 2A and S1C). Free fatty acids (FFAs) were elevated significantly in both PKO and DKO livers but not in sera (Figure S1D). Conversely, cholesterol increased significantly in both PKO and DKO sera but not in their livers (Figure S1E). These data demonstrate that both PKO and DKO mice developed hepatosteatosis, and that additional Shp2 ablation modestly aggravates the phenotype induced by Pten loss.

To probe the underlying mechanism, we profiled the expression of key genes involved in lipid metabolism and gluconeogen-esis in 2-month-old livers (Figure 2B). Pparg expression was induced more robustly in livers of DKO and PKO than WT and SKO, and Cd36, a Pparg downstream target that regulates lipid uptake and transport, was significantly upregulated in PKO and even more in DKO, suggesting increased lipid uptake in the livers. Increase of Fas and Scidl, encoding two critical lipogenic enzymes, was also detected in DKO and PKO, suggesting enhanced de novo lipid synthesis. Hsl and Atgl, coding for rate-limiting enzymes in lipolysis, were modestly or significantly lower in DKO or PKO livers, providing evidence of impaired hepatic lipolysis. Of special interest, we found that the expression of genes involved in lipid oxidation, including Ppara, Aco, Lead, and Mcad and genes involved in gluconeogenesis, such as G6p and Pepck, were significantly attenuated in DKO but not in PKO liver. We also examined the expression profiles of the same set of genes in 1-month-old livers, and obtained consistent results in 1- and 2-month-old mice of the four genotypes (Figure S1F). These data suggest that the severe hepatos-teatosis phenotype in DKO was caused collectively by metabolic

HCC-ICC

¿III É i v p№1 iiSSWfis ml I jggfós , , ,

((№ " SffiS 'i mm » ill® listel ¡P¡Í¡1 In fea llii. ISBS

X. - p f&v&fl 'k-t.-Z 11 Éllllll ISlllSli KK-g^lgjii RBh8k»5S itS^SeS

Max Tumor Sizes (mm)

■ ■ ■□

a "a Cfl S O O O

Tumor number/mouse

1\J CJl mot

Tumor ®

incidences (%)

Liver/ body weight ratios

Liver weight (g)

to W J* o

Metastasis No Metastasis

KSHBÉSsS (Mp

HCC-ICC

■ ill! ^^HreC i (. WM

mm i^mMt felpáis Mal»íí IVSMPS 1 ll ¿I^fifii r, .

Om m z

Figure 2. Dual Deletion of Shp2 and Pten in Hepatocytes Promotes Hepatic Steatosis

(A) Left: representative H&E and Oil-Red-O staining of liver sections. Right: quantification of percentage of positive Oil-Red-O staining area (n = 3).

(B) mRNA levels of key genes in lipid metabolism and gluconeogenesis were determined with qRT-PCR and compared (n = 5).

All samples used in this figure were obtained from 2-month-old mice. Data are shown as means ± SD. *, **, or*** indicates SKO, PKO, or DKO versus WT. $, $$, or $$$ indicates DKO versus SKO. #, ##, or ### indicates DKO versus PKO. *, $, or # indicates p < 0.05. **, $$, or ## indicates p < 0.01. ***, $$$, or ### indicates p< 0.001. See also Figure S1.

changes in lipogenesis, uptake, lipolysis, oxidation, and gluconeogenesis. After feeding the mice high-fat diet (HFD) for 1 month, we observed significantly elevated triglyceride levels in DKO mice on HFD relative to normal chow diet, but not in WT controls (Figure S1G), suggesting increased susceptibility of DKO mice to HFD-induced hepatic steatosis.

Histological examination revealed severe infiltration of inflammatory cells in portal triads with extension to the parenchyma in DKO liver (Figure 3A), and mild inflammation was observed in SKO, but not in PKO, consistent with previous reports (Bard-Chapeau et al., 2011; Horie et al., 2004). Immunostaining showed that Kupffer cells (F4/80+), neutrophils (Ly6G+), and T cells (CD3+), but not B cells (B220+), were significantly increased in DKO liver (Figure 3A). Marked increase in the spleen-to-body weight ratios was detected in SKO, and the ra-

tios were even more elevated in DKO mice, indicating enhanced systematic inflammatory and immune responses (Figure 3B). Consistently, microarray and qRT-PCR analyses demonstrated that many inflammatory factors and related receptors, such as Tnfa, Fgf21, Pdgfa, Pdgfc, Ccl21a, and Cx3cl1, were more dramatically induced in DKO than the other three groups (Figures 3C and S2D).

Moreover, DKO livers showed the most severe fibrosis, as evaluated by Picro-Sirius Red and Masson'sTrichrome staining (Figure 3D). Activated hepatic stellate cells, the major source of collagen, were largely expanded in DKO, as determined by a-SMA staining (Figure 3D). TUNEL staining showed more hepa-tocyte apoptosis in DKO livers (Figure 3E), accompanied by drastic increase in serum ALT levels (Figure 3F), further indicating more severe hepatic injuries. In addition, ballooning

Figure 1. Shp2 and Pten Cooperate to Suppress Liver Tumorigenesis

(A) Representative macroscopic view of livers.

(B) Tumor incidences, maximal tumor diameters, tumor number/mouse, liver weight, and liver/body weight (BW) ratios were compared (n = 5-19). Data are represented as mean ± SD. #p< 0.05; ##p< 0.01; and ###p< 0.001, for DKO versus PKO. p value < 0.001 in all comparisons for PKO versus WT. p value <0.001 in all comparisons for DKO versus WT or DKO versus SKO.

(C) Representative H&E staining of hepatocellular adenoma (HCA), hepatocellular carcinoma (HCC), intrahepatic cholangiocarcinoma (ICC), and mixed HCC-ICC tumors.

(D) Representative H&E and immunostaining of tumor sections from DKO mice.

(E) H&E and immunostaining of lung sections from DKO mice.

(F) PCR analysis (upper) of genomic DNAs with Shp2 deletion (Shp2D), Pten deletion (PtenD) and Cre, and immunoblot analysis (lower) of expression of cell cycle-related genes, oncogenes, Shp2 and Pten in liver tissues or tumors from the four groups of mice.

See also Table S1.

Figure 3. DKO Mice Develop Early-Onset NASH

(A) Top: representative H&E and immunostaining of immune cells In the livers. Bottom: quantification of B220, Ly6G, CD3-posltive cells and percentage of positive F4/80 staining area (n = 3).

(B) Spleen/body weight ratios were compared (n = 6).

(C) Heat map with liver cDNA microarray data show mRNA levels of secretory factors and related receptors with color intensity. Each row indicates one mouse. Color scale is provided as reference.

(D) Top: Picro-Sirius Red, Masson's Trichrome, and aSMA immunostaining of liver sections. Bottom: quantification of percentage of positive Sirius Red, Tri-chrome, and a-SMA staining area (n = 3).

(E) Top: TUNEL assay, Ki67 and PCNA immunostaining of liver sections. Bottom: quantification of TUNEL, Ki67 or PCNA-positive hepatocytes (n = 3-6).

(F) Serum ALT was measured and compared (n = 6).

(legend continued on next page)

hepatocytes and Mallory-Denk bodies (MDBs) were detected in DKO livers (Figures 3G and 3H). PCNA and Ki67 staining showed more active hepatocyte proliferation (Figure 3E). Consistent with our previous report (Li et al., 2014), hepatic bile acid levels were elevated in SKO, but the increase was compromised by additional Pten deletion in DKO livers (Figure S2A). Accordingly, Cyp7a1, a key enzyme for bile acid synthesis, was induced in SKO, but this induction was blunted in DKO livers (Figures S2B and S2C). We further analyzed the toxicogenomic changes using IPA-Tox with the microarray data. Dual deletion of Shp2 and Pten induced changes related to NASH, such as liver fibrosis, steatosis, cell death and proliferation, inflammation, hepatitis, and enlargement, as shown at the top of the list (Figure S2E). In conclusion, young DKO animals exhibited typical NASH signs, resembling a subset of human NASH patients (Chalasani et al., 2012).

Next, we deleted Shp2 and Pten acutely in adult mice (Shp2fl/fl:Ptenfl/fl) with adeno-associated virus (AAV) expressing Cre (AAV-Cre), with AAV-green fluorescent protein (GFP) as a control, and then subjected the mice to CCl4 treatment. Immuno-blotting showed that Shp2 and Pten were efficiently removed by AAV-Cre infection (Figure S2F). CCl4 treatment led to mild liver damage in control mice but caused severe damage in livers with both Shp2 and Pten deleted, as revealed by increased fibrosis and inflammation, serum ALT, and spleen-to-body weight ratios (Figures S2G-S2J). These results suggest a cooperative role of Shp2 and Pten in protection against hepatic injuries.

The Liver Tumor-Initiating Cell Population Expands in Young DKO Mice

The earlier detection of tumor nodules in DKO than PKO mice suggests that additional Shp2 deletion accelerates the initiation and/or progression of liver tumorigenesis triggered by Pten loss. To sort out these possibilities, we evaluated the expression of biomarkers for liver tumor-initiating cells (TICs) in young mice at 2 months of age, when no tumor nodules were visible. qRT-PCR detected significantly elevated mRNA levels of EpCam, Ck19, Prominin1, Cd44, and Ly6D, but not Afp, in DKO livers (Figure 4A). Consistently, immunostaining detected increased number of cells positive for EpCAM, Ly6D, CD44, CK19, and CD44v6 in DKO liver sections (Figure 4B), and A6+ oval cells were also increased (Figure 4B). Although Sox9 mRNA was not changed (Figure 4A), the SOX9+ cell population was dramatically expanded in DKO liver (Figure 4B). Of note, cells positive for EpCAM, CK19, CD44v6, or SOX9 were mainly located in the portal triad area, whereas those stained with CD44 or Ly6D were scattered in the liver, suggesting heterogeneity of the TIC population. Furthermore, co-immunostaining for CD44v6/SOX9 and EpCAM/SOX9 consistently detected the expansion of TICs in DKO livers (Figure 4C). We then examined tumors and surrounding tissues in 12-month-old DKO mice, using SOX9 as a marker

(Figure 4D). Very few SOX9+ cells were detected in non-tumor tissues or HCAs, but extensive staining of SOX9 was observed in ICCs, HCCs, mixed ICC/HCC tumors, and lung-metastasized nodules.

Following examination of physical markers, we performed functional analysis of liver TICs in young mice, before detection of visible tumor nodules. We isolated hepatocytes from 2- to 3-month-old mouse livers and injected 1.2 x 105 viable cells intra-splenically into 21-day-old MUP-uPA transgenic mice, as described previously (He et al., 2013). By examining liver tumorigenesis in recipients after 5 months, we found no tumors in mice transplanted with WT or SKO hepatocytes, and tiny tumors in two of seven recipients for PKO hepatocytes. In contrast, seven of eight mice that received DKO hepatocytes had tumors and their tumor burdens were much heavier (Figures 4E and 4F). Pathological examination detected two types of tumor, HCC and HCC/ICC mixture in recipients for DKO hepatocytes, and both types of tumor contained SOX9+ cells (Figure S3A). PCR analysis of genomic DNA and immunoblotting of proteins extracted from these tumors confirmed deletion of Shp2 and Pten, proving that these tumors originated from transplanted donor cells (Figure 4G). Tumor nodules were even detected in the lung of two recipients of DKO hepatocytes (Figure 4H), which were SOX9+ and histologically similar to HCCs (Figure S3B). Thus, concurrent Shp2 and Pten deletion induced earlier expansion of a functional liver TIC population.

c-jun Is Upregulated in Shp2/Pten-Deficient Liver

We investigated the molecular mechanisms that drive the expansion of TICs in DKO liver. Global gene expression profiling by cDNA microarray analysis of 2-month-old liver samples revealed upregulation of several proto-oncogenes in DKO mice (Table S2). In particular, on top of the list is c-jun, a proto-oncogene that plays a critical role in hepatogenesis, liver regeneration, tumor initiation, and progression (Eferl and Wagner, 2003; Ran-sone and Verma, 1990).

c-jun mRNA increase detected in the microarray analysis was confirmed by qRT-PCR (Figure 5A). Immunoblotting showed that both the protein amount and phosphorylation of c-Jun were elevated in DKO livers (Figure 5B), indicating higher expression and activity. Consistently, immunostaining detected more c-Jun+ cells in DKO liver sections (Figure 5C). c-Jun upregulation could be stimulated by altered intracellular signaling and hepatic microenvironment. Indeed, both liver and systematic inflammations were significantly enhanced in DKO animals (Figures 3A and 3B). Multiple cytokines, growth factors, and related receptors were upregulated in DKO livers (Figures 3C and S2D). We examined reactive oxygen species (ROS) and observed more extensive staining with dihydroethidine (DHE) for superoxide anions, and with trans-4-hydroxy-2-nonenal (4-HNE) for lipid peroxidation aggregates in DKO livers (Figure 5D). Interestingly, the periportal areas, where more TICs were found, showed

(G and H) Representative H&E staining shows Mallory-Denk body and ballooning hepatocytes (arrowhead) in DKO liver.

All samples used in this figure were obtained from 2-month-old mice. Data are shown as means ± SD. *, **, or*** indicates SKO, PKO, or DKO versus WT. $, $$, or $$$ indicates DKO versus SKO. # or ### indicates DKO versus PKO. *, $, or # indicates p < 0.05. ** or $$ indicates p < 0.01. ***, $$$, or ### indicates p < 0.001. See also Figure S2.

>1 ¿1 jH K . ' . ■> ¿ - 200llM

j '•s: * «, -í , "V . '* 200|JM

il 100MM

i 'SISSÎm ■HHHBBBKEH99H3H 111 , 1 -■ SWàêmÊMimm Hr- ■ ' ■ $m

■ mm mmm ¡SR

< i "s"*- ' v< l' fftf"^ " gjÉ • , ' ' - \ V-- ¿ ■ ,r • - -7 : „ B" ^ ' JOOpflU * . ' MM

D >■ tn 3 »g 50MM

m ¡¡¡¿1 NT «S HCA '102f ' - V ' . ' . : HCC :

. . if ' ' ; ip? >: 3 ' HÇQ/ipo,<<a¡¡

-m ijpixpg Sgl| B î i Lung without metastasis 2<K Lung with ■ metastasis " . Ï/I' Mf¿. A ti'j' ' . • 50*pM ■ MM

Donor: WT

1 à** , ., * V

5 (•' '•'J-V 4 - ** V • ~ V * SffM

+ 150 >< -a

□ wt

■ sko

■ pko

■ dko

„1001

^ 80 </> S 60 c !5 40 £ 20 I Max Tumor sizes(mm)

-APten

-AShp2

- * *1 -UPA

ii -Shp2

— -Pten

----— -GAPDH

livers tumors

w „ g

i» h -S

■ sko

■ pko

■ dko

(legend on next page)

increased staining for the ROS markers. Consistently, we detected more severe DNA damage in DKO livers, by staining with pH2AX antibody (Figure 5D). qRT-PCR analysis showed lower expression of H2O2 scavengers Gpx1, Gpx2, and Catalase and antioxidant enzymes Sod1, Sod2, Gsta1, Gst, and Gstp1 in SKO and DKO livers (Figure 5E).

To evaluate a possible microenvironmental effect on c-Jun up-regulation, we treated DKO mice with lenalidomide, an analog of thalidomide that acts to inhibit expression of cytokines, such as tumor necrosis factor a (TNF-a) (Zhu et al., 2014). Increased TNF-a was detected in DKO livers (Figure S2D), and lenalidomide treatment downregulated TNF-a expression (Figure 5F). Consistently, the mRNA, protein, and phosphorylation levels of c-Jun also decreased (Figures 5F and 5G), accompanied by reduced TIC population (Figure S4). These results suggest that the inflammatory factors, especially TNF-a, contributed to c-Jun upregulation in DKO livers. However, lenalidomide treatment did not suppress c-Jun to the basal level, suggesting that intra-cellular signaling altered by Shp2 and Pten deletion may also play a role in c-Jun upregulation. As expected, p-Akt levels were markedly elevated by Pten loss, and this effect was modestly attenuated by additional Shp2 deletion (Figure 5H). Decreased p-Erk signals were observed in SKO and DKO livers, relative to WT and PKO, reinforcing a positive role of Shp2 in Erk activation (Lai et al., 2004; Neel et al., 2003), without an obvious effect on p38 activity. In contrast, p-Jnk signals were robustly induced by ablating Shp2 in SKO and DKO livers (Figure 5H), supporting our previous data indicating a negative role of Shp2 in Jnk activation (Bard-Chapeau et al., 2006; Shi et al., 1998). pY-Stat3 was elevated in SKO livers as previously reported (Bard-Chapeau et al., 2011), but this effect was compromised by additional Pten deletion in DKO livers (Figure 5H). As it has been well documented that Akt can activate c-Jun/AP-1 (Cross et al., 1995; Gotschel et al., 2008; Li et al., 2004; Peloponese and Jeang, 2006), we believe that the excessive c-Jun expression and activation in DKO liver are likely contributed by simultaneous over-activation of Akt and Jnk induced by removal of Pten and Shp2, respectively, together with the elevated inflammation and ROS levels in the hepatic microenvironment (Figure 7E).

Inhibition of c-Jun Suppresses Liver Tumorigenesis but Not NASH

We investigated the potential role of c-Jun in TIC expansion by examining its expression profile. Co-immunostaining showed

that almost all EpCAM+ or A6+ cells were also positive for c-Jun in 2-month-old DKO liver (Figure 6A), confirming c-Jun induction in TICs. In cultured HepG2 cells, the subpopulation with higher EpCAM also exhibited increased c-Jun expression (Figure 6B). We then determined the effect of c-Jun inhibition by expressing TAM67, a dominant-negative form of c-Jun acting to sequester c-Jun into low-activity complexes (Brown et al., 1993; Shimizu et al., 2008). In an in vitro sphere-forming assay (He et al., 2010), expressing TAM67 in HepG2 cells significantly decreased the sphere numbers, compared to the control (Figure 6C). We then introduced TAM67 into 1-month-old SKO, PKO, and DKO liver using AAV8 virus, and evaluated TIC expansion and tumorigenesis. qRT-PCR analysis of the downstream targets of c-jun, c-fos, JunB, and JunD (Eferl and Wagner, 2003) showed that TAM67 potently suppressed c-jun and slightly reduced c-fos, without obvious effect on JunB or JunD in DKO livers (Figure S5A). After 1 month of infection, the TIC population was severely suppressed in DKO mice while only slightly decreased in SKO and PKO mice (Figures S5B and S5C). The expression of TAM67 was high at 1 month after infection, but barely detectable after 6 months (Figure S5D). TAM67 delivery significantly suppressed the tumor number and moderately reduced tumor sizes, without affecting tumor incidences and liver weights, compared to GFP controls (Figures 6D and 6E). qRT-PCR and immunostaining also showed suppression of TIC markers including EpCAM, CK19, Ly6D, CD44, and SOX9 in DKO livers by expressing TAM67 (Figures 6F and 6G). Notably, the NASH symptoms were not improved by inhibiting c-Jun in DKO livers (Figures S5E-S5G), which indicates its direct effect on liver tumorigenesis by inhibiting TIC expansion, independent of steatohepatitis. A previous report showed that c-Jun enhanced DEN-induced liver tumorigenesis by inducing Sur-vivin (Min et al., 2012). By co-immunostaining, we detected co-localization of Survivin and EpCAM in DKO liver sections (Figure 6H), and expressing TAM67 decreased Survivin+ cells in DKO liver (Figure S5H), supporting a role of Survivin downstream of c-Jun.

SHP2 and PTEN Deficiencies Correlate with Poor Prognosis in HCC Patients

To assess the clinical significance of SHP2 and PTEN deficiency, we performed immunostaining for SHP2 and PTEN on a tissue microarray (TMA) with 350 self-paired human HCC specimens (Tables S3, S4, and S5). Both SHP2 and PTEN showed

Figure 4. Early Expansion of Liver TICs in DKO Mice

(A) mRNA levels of TIC markers were determined with qPCR and compared among 2-month-old WT mice (n = 5).

(B) Left: representative immunostaining of TIC markers on liver sections from 2-month-old mice. Right: quantification of EpCAM, A6, Ly6D, CD44, CK-19, SOX9, and CD44v6-positive cells (n = 3).

(C) Top: representative co-immunostaining of CD44v6/SOX9 and EpCAM/SOX9 on liver sections from 2-month-old mice. Bottom: quantification of CD44v6/ SOX9 and EpCAM/SOX9-positive cells (n = 3).

(D) Representative SOX9 immunostaining of sections from 12-month-old DKO mice.

(E-H) Data from MUP-UPA mice 5 months after transplantation with DKO hepatocytes. (E) Macroscopic view of representative livers from MUP-uPA mice 5 months after transplantation with hepatocytes from 2- to 3-month-old WT, SKO, PKO, and DKO. (F) Tumor incidences, sizes (maximal tumor diameters), and tumor numbers/mouse were compared (n = 7~9). Data are represented as mean ± SD. (G) PCR analysis (upper) of genomic DNA and immunoblot analysis (lower) of lysates from respective liver or tumor tissues. (H) Macroscopic view of a representative lung with metastases.

Data are shown as means ± SD. *, **, or *** indicates SKO, PKO, or DKO versus WT. $ indicates DKO versus SKO. # or ## indicates DKO versus PKO. *, $, or # indicates p < 0.05. ** or ## indicates p < 0.01. *** indicates p < 0.001. See also Figure S3.

Figure 5. Elevated Expression and Activation of c-Jun in DKO Liver

(A) mRNA levels of c-jun were determined by cDNA mlcroarray (left) and qPCR (right) (n = 5).

(B) Protein and phosphorylation levels of c-Jun were determined by immunoblot.

(C) Left: c-Jun immunostaining of liver sections. Right: quantification of c-Jun-positive cells (n = 3).

(D) Top: DHE and immunostaining for 4-HNE and pH2AX were performed on liver sections. Bottom: quantification of 4-HNE, pH2AX-positive cells, and percentage of positive DHE staining area (n = 3).

(E) mRNA levels of key genes in reactive oxygen species (ROS) scavenging were determined by qPCR (n = 5).

(F) mRNA levels of TNF-a and c-Jun were determined by qPCR (n = 4). Data are shown as means ± SD. * indicates p < 0.05. (Lena: Lenalidomide.)

(G) Protein and phosphorylation levels of c-Jun were determined by immunoblot.

(legend continued on next page)

Figure 6. c-Jun Is Critical for TIC Expansion and Tumorigenesis in DKO Mice

(A) Co-immunostaining for A6/c-Jun, EpCAM/c-Jun on liver sections from 2-month-old DKO mice.

(B) HepG2 cells were FACS sorted for EpCAM expression. c-Jun and EpCAM protein levels were determined by immunoblotting.

(C) Microscopic view of representative hepatospheres formed by HepG2 cells infected with AAV-GFP or AAV-TAM67. Bar chart shows the quantification of hepatospheres/1,000 cells. Experiment was performed three times with three wells/group each time.

(D) Macroscopic view of representative livers from DKO infected with AAV-GFP or AAV-TAM67.

(E) Tumor number/mouse, maximal tumor diameters, liver weight, liver/body weight (BW) ratios and tumor incidences were determined and compared between mice infected with AAV-GFP or AAV-TAM67 (n = 6~7).

(F) mRNA levels of TIC markers were determined by qPCR and compared between mice infected with AAV-GFP or AAV-TAM67 (n = 5).

(G) Left: immunostaining for EpCAM and SOX9 was performed on liver sections from mice infected with AAV-GFP or AAV-TAM67. Right: quantification of SOX9 and EpCAM-positive cells (n = 6~7).

(H) Co-immunostaining for EpCAM and Survivin was performed on DKO liver sections. Data are presented as mean ± SD (*p < 0.05, **p < 0.01, ***p < 0.001).

See also Figure S5.

cytoplasmic staining, and scores of 0-3 were given, with 0 indicating no or very weak, 1 forweak,2for medium, and 3 for strong staining (Figure S6B). We compared SHP2 or PTEN expression in each tumor with the paired tumor-surrounding tissue. In total, we identified 106, 113, and 131 pairs of specimens with SHP2 expression higher, comparable and lower in tumors than surrounding tissues, respectively, whereas the corresponding numbersfor PTEN were 58,86, and 191, respectively (Figure 7A).

Similar results of variable SHP2 and PTEN expression were obtained by examining a different TMA that contained 45 pairs of human HCC tumors and surrounding tissues purchased from US Biomax (Figure S6A).

We then assessed the prognostic value of SHP2 and PTEN expression, and found that lower expression of SHP2 or PTEN alone in tumors was significantly associated with shorter overall and disease-free survival time (Figures 7B and 7C). Deficient

(H) Analysis of signaling pathways with immunoblotting of liver lysates.

All samples used in (A)-(E) and (H) were from 2-month-old mice. Data are shown as means ± SD. *, **, or *** indicates SKO, PKO, or DKO versus WT. $ or $$ indicates DKO versus SKO. # or ## indicates DKO versus PKO. *, $, or # indicates p < 0.05. **, $$, or ## indicates p < 0.01. *** indicates p < 0.001. See also Figure S4 and Table S2.

Figure 7. SHP2 and PTEN Deficiencies in Human HCCs

(A) Expression of SHP2 and PTEN was analyzed by immunostaining and compared in 350 pairs of human HCC and tumor-surrounding tissue samples (T: tumor; S: tumor-surrounding tissue).

(B-D) Kaplan-Meier survival curves of overall survival (left) and disease-free survival (right) based on SHP2 (B), PTEN (C), and Shp2 and Pten (D) expression levels in HCC tissues. Samples were grouped by low (score % 1) or high (score > 1) expression of each protein. Log-rank test was performed, and p values are shown. (E) A model for enhanced liver tumorigenesis driven by combined deletion of Pten and Shp2. Pten loss leads to Akt over-activation and Shp2 deficiency promotes Jnk activation, resulting in increased c-Jun expression and activation. Ablating Pten and Shp2 also leads to elevated ROS, inflammation, and steatohepatitis, etc., driving earlier genesis of TICs and enhanced HCC development. See also Figures S6 and S7 and Tables S3, S4, S5, and S6.

SHP2 expression in tumors also correlated with larger tumor sizes and higher serum AFP levels (Table S3). To determine the prognostic value of combined SHP2 and PTEN deficiencies, we further classified the 335 HCC patients into four groups: I, low expression of both SHP2 and PTEN; II, low SHP2 but high

PTEN expression; III, high SHP2 but low PTEN expression; and IV, high expression of both SHP2 and PTEN. SHP2L or PTENL included SHP2 or PTEN staining scores of 0-1, and SHP2H or PTENH were for staining scores of 1-3 (excluding 1). The group I patients with low expression of both showed significantly

shorter overall and disease-free survival time (Figure 7D). Furthermore, the tumor sizes and AFP amounts of group I patients were significantly larger and higher, respectively (Table S5). These data argue for the clinical relevance of SHP2 and PTEN deficiencies in human HCC development and prognosis.

We further examined SHP2 and PTEN expression in liver samples of patients at NAFLD, NASH12, or NASH34 stages (Table S6). Immunostaining detected variable levels of SHP2 and PTEN expression; the SHP2 expression correlated inversely with the advancing disease stages, whereas PTEN expression did not show significant difference among these groups (Figures S7A and S7B). However, two recent papers reported inverse correlation of PTEN expression in hepatocytes with NAFLD progression (He et al., 2016; Sanchez-Pareja et al., 2016). Notably, the expansion of SOX9+ cells was associated with the advancement of NAFLD to NASH stages (Figure S7C). By examining SOX9 expression in the TMA with 335 HCC specimens, we identified 42 samples with low SHP2 and PTEN expression and also with large number of SOX9+ cells (Figure S7D). Together, these results suggest involvement of SHP2 and PTEN deficiencies in human pathogenesis of NASH-driven HCC.

DISCUSSION

The concerted actions of Shp2 and Pten in suppressing liver tumorigenesis is astonishing, because we have just reported mutually neutralizing effects of concurrent Shp2 and Pten deletion in myeloproliferation and leukemogenesis (Zhu et al., 2015). The hepatic phenotype of DKO mice unequivocally defines a tumor-inhibitory role of Shp2 in the liver, by working in concert with a classical tumor suppressor Pten in hepatocytes. A model to illustrate the underlying mechanisms is presented in Figure 7E. Pten loss leads to Akt over-activation and Shp2 deficiency promotes Jnk activation, resulting in increased c-Jun expression and activation. Ablating Pten and Shp2 also leads to elevated ROS, inflammation and steatohepatitis, driving earlier genesis of TICs and enhanced tumorigenesis.

NAFLD has become a serious public health problem worldwide (Michelotti et al., 2013), and NASH, an advanced state of NAFLD, is characterized by hepatic steatosis, injury, and inflammation with or without fibrosis (Chalasani et al., 2012). The pathogenesis from NAFLD to NASH likely initiates with metabolic disorders that cause hepatosteatosis, and then progresses to steatohepatitis, due to oxidative stress and inflammation, etc. Pten deficiency in hepatocytes induced excessive activation of the PI3K/AKT pathway, increased lipogenesis, and reduced lipolysis, resulting in hepatic lipid accumulation and injuries. Additional removal of Shp2 did not dramatically exacerbate the fatty liver phenotype but did cause liver inflammation, damage, and fibrosis at young age. Thus, combined Pten and Shp2 deficiency induced NASH immediately without even a NAFLD stage, by simultaneously triggering multiple hits in DKO mice.

The molecular mechanisms underlying NASH-induced liver carcinogenesis remain to be elucidated, and several mouse models have recently been developed to tackle this problem. Mice fed with CDE (choline deficiency and ethionine supplemented) diet developed fatty liver in 14 days, and long-term

(19-month) feeding of this diet caused HCC (Aharoni-Simon et al., 2011). However, CDE diet is highly toxic and led to high mortality with long-term application, which limits its use as a NASH-HCC model (Yoshida et al., 1993). Mice treated with DEN and fed with HFD developed HCC at 50 weeks (Park et al., 2010), but it is unclear whether these mice had NASH symptoms at young age. Feeding MUP-uPA transgenic mice with HFD induced signs of NASH at young age and HCC development at late stages (Nakagawa et al., 2014). Another NASH-HCC model was established by feeding mice with acholine-defi-cient HFD, which recapitulated human NASH characteristics, and 25% of the mice developed HCC after 1 year of feeding (Wolf et al., 2014). By deleting Shp2 and Pten, we have established a NASH-HCC model, in which HCC develops spontaneously in 7 months, without DEN treatment. The DKO mice, recapitulating many features of NASH patients, may represent a useful animal model for extensive research of molecular signals that drive NASH to HCC.

The earlier genesis of TICs evidently contributed to the severe phenotype of liver tumorigenesis in the DKO mouse. This conclusion is not only based on detection of physical markers but also relies on the functional determination of TICs in vivo. Intra-splenic transplantation of hepatocytes isolated from 2-month-old DKO liver into MUP-uPA mice caused growth of numerous tumor nodules in the liver recipients. We believe that the early-onset NASH is a critical factor for TIC activation and expansion. Excessive lipid accumulation in hepatocytes is toxic and boosts ROS levels, causing genome instability and cell death. These damages could switch on the innate immune system, represented by expansion of Kupffer cells, infiltration of inflammatory cells, and secretion of cytokines and chemokines. Moreover, liver injuries could also activate hepatic stellate cells (HSCs) to synthesize and secrete more collagens that induce fibrosis. Together, these hepatic micro-environmental factors promoted TIC activation.

One molecular link between the micro-environment and activating TICs is possibly c-Jun, which was potently induced and activated by inflammatory factors and ROS in DKO liver combined with augmented Akt and Jnk signaling (Figure 7E). c-Jun expression overlapped with progenitor cell markers EpCAM and A6, and inhibiting c-Jun with a dominant-negative form, TAM67, in HepG2 cells impaired the self-renewal capacity in vitro. Expression of TAM67 in DKO liver also suppressed TIC expansion and tumorigenesis in vivo, even though the NASH symptoms were not improved. Therefore, c-Jun inhibition uncouples the micro-environmental signals from TIC activation. Of note, c-Jun is not likely the sole link between these processes, and other mechanisms may be involved in over-activation of TICs.

By screening two independent tissue microarrays of human liver cancer (mainly HCC) samples, we detected deficient expression of SHP2 and/or PTEN in tumors relative to the surrounding tissues. Strikingly, the survival curves indicated poor prognosis of the patients with SHP2 and PTEN deficiencies. In a recent report (Jiang et al., 2012), the authors screened 333 human HCC samples and detected significantly decreased expression of SHP2 in 235 tumors in correlation with poor prognosis. Analysis of a different cohort of HCC samples detected

overexpression of SHP2 in association with HCC progression and poor prognosis (Han et al., 2015). The complexity of big patients' data are expected and likely represents the diverse age, sex, and disease backgrounds, compared to the mouse models with clean genetic background and targeted deletion of specific genes.

In contrast to the anti-tumorigenic role of Shp2 described here, a recent report indicated that inhibiting Shp2 by short hairpin RNA (shRNA) or chemical compound selectively suppressed proliferation of cancer cells driven by receptor tyrosine kinases (Chen et al., 2016). Although hyperactivation of Akt is frequently detected in human cancers and is therefore a therapeutic target, hepatic deletion of Akt1 in Akt2_/~ mice surprisingly triggered spontaneous HCC development (Wang et al., 2016). Given the bidirectional roles of these signaling molecules in liver tumorigenesis and the disappointing therapeutic effect of Sorafenib, we propose here that targeting further downstream and signal-culminating molecules may be a more efficient pharmaceutical means for treatment of liver cancer.

EXPERIMENTAL PROCEDURES Animal Protocols

Hepatocyte-specific Shp2 KO mice (SKO, Shp2hep~'~) were generated and characterized as described previously (Bard-Chapeau et al., 2006, 2011; Li et al., 2014). Pten"ox"lox (Lesche et al., 2002) mice were bred with Albumin-Cre+ transgenic mice to generate hepatocyte-specific Pten KO mice (PKO) and crossed with Shp2flox/flox to produce Shp2 and Pten double-knockout (DKO) animals. MUP-uPA transgenic mice were kindly provided by M. Karin (UCSD). The animal protocols (s09108) for all procedures were approved by the UCSD Institutional Animal Care and Use Committee (IACUC). Collection of mouse liver and other samples was done at the same time during the day. qPCR, H&E staining, immunostaining, immunoblotting, and fluorescence-activated cell sorting (FACS) were performed following standard procedures, and ImageJ was used to quantify staining data. PCR primers and antibodies are listed in Table S7. Primary hepatocytes were isolated from 2- to 3-month-old mice as described previously (Bard-Chapeau et al., 2011). Mice were fed with HFD (Research Diets; catalog no. D12451). Mice were injected intraperi-toneally (i.p.) with olive oil orCCl4(1 ml/kg body weight; Sigma-Aldrich; catalog no. 289116) twice a week for 4 weeks. Mice were oral gavaged with lenalido-mide (50 mg/kg per day for 5 days; Selleck Chemicals; catalog no. S1029). Intrasplenic transplantation was performed as described previously (He et al., 2010, 2013). Briefly, 1.2 x 105 viable hepatocytes in 30 mL of PBS were transplanted into 21-day-old MUP-uPA mice via intrasplenic injection with a 26G needle. Five months after transplantation, mice were sacrificed for examination of liver tumors. Liver and/or serum triglycerides, bile acids, cholesterol, and FFA levels were determined with commercial kits (triglyceride and cholesterol: Sekisui Diagnostics; catalog no. 236-60,234-60; bile acid: Diazyme Laboratories; catalog no. DZ042A-K; FFA: Cayman Chemical; catalog no. 700310). Oil-Red-O, Picro-Sirius Red, and Masson's Trichrome staining were performed with commercial kits (American MasterTech; catalog nos. STORO, STPSR, and KTMTR).

Microarray, Heat Map, and IPA Analysis of Gene Expression

Microarray analysis of RNAs was performed as previously described (Li et al., 2014). Heat map was generated using GENE-E (https://www.broadinstitute. org/cancer/software/GENE-E/index.html) with microarray data, which was also analyzed with Ingenuity IPA (http://www.qiagenbioinformatics.com/ products/ingenuity-pathway-analysis/).

Cell Culture and Hepatosphere-Forming Assay

HepG2 (ATCC HB-8065) cells were cultured following ATCC instructions. Hepatosphere-forming assay was performed as previously described

(He et al., 2010), with modification. Briefly, 5,000 cells were seeded in six-well ultra-low attachment plate (Corning 3471) with sphere formation medium (DMEM/F12 medium, 20 ng/mL EGF, 20 ng/mL bFGF, 1% B27 supplement). Spheres were counted under microscope after 7-day culture.

DHE Staining

DHE staining was performed as previously described (Maeda et al., 2005). Briefly, frozen liver sections were stained with 2 mM dihydroethidine hydrochloride (Life Technologies; catalog no. D11347) for 30 min at 37°C.

Generation and Administration of Recombinant AAV8 Viruses

The AAV expression system (Stratagene; catalog nos. 240071 and 240074) was used to produce AAV- GFP and AAV-TAM67 virus, with pAAV8-RC replacing pAAV-RC. Primers used for AAV-TAM67 construction and virus titration are listed in Table S7. AAV viruses (~4 x 1011 GV/mouse) were diluted in PBS and injected through the tail vein. AAV-Cre virus was purchased from the Vector Core of University of Pennsylvania.

Human Materials

All archival specimens for human HCC Tissue Microarray were obtained from 350 consecutive patients after surgical resection in 2007 with pathologically confirmed HCC (Tables S3, S4, and S5). The experimental protocols described in this study complied with the Ethics Review Committee of Zhongshan Hospital, Shanghai, China, and every patient provided written informed consent before enrollment. TMAs were constructed as previously described (Liao et al., 2012). All patients were monitored postoperatively until January 2012. All of the immunohistochemistry (IHC) staining and data analysis were done at UCSD. Kaplan-Meier survival analysis was performed with GraphPad Prism 5 (GraphPad Software). Commercial TMA was purchased from US Biomax (catalog no. LV1504). Human NAFLD, NASH12, and NASH34 patients' liver samples were collected at UCSD Medical Center (Table S6). The experimental protocol described in this study was approved by the human subjects institutional review board at UCSD, and all patients provided written informed consent at the initial visit.

Statistical Analyses

Two-tailed Student's t test, one-way ANOVA, c2 tests were performed with R (version 3.1.0). p < 0.05 was considered as statistically significant.

ACCESSION NUMBERS

The accession number for the microarray data for liver samples of 2-month-old mice utilized in this study is GEO: GSE74732.

SUPPLEMENTAL INFORMATION

Supplemental Information includes seven figures and seven tables and can be found with this article online at http://dx.doi.org/10.1016/j.celrep.2016.11.048.

AUTHOR CONTRIBUTIONS

X.L. and G.-S.F. conceived and designed the project, analyzed the data, and wrote the paper. G.-S.F. provided the reagents. X.L. designed and performed most of the experiments. X.W. analyzed the expression of SHP2 and PTEN in HCC patients with IHC and western blot. R. Liao, S.-J.Q., and J.F. provided human HCC TMA samples. C.H. and R. Loomba provided human NAFL and NASH liver sections. N.M.V. helped pathology analysis. X.L. and N.A. purified the AAV8 viruses. S.L. and H.H.Z. helped analyze the data. K.L.H. and K.N.M. helped with the human HCC TMA staining and data analysis. K.L.H., C.C., N.A., and K.N.M. helped with genotyping, immunoblotting, and immunostaining.

ACKNOWLEDGMENTS

We thank Drs. D. Brenner, M. Karin, and other colleagues for advice and reagents. A6 antibody was a generous gift from V.M. Factor (NIH). This work

was supported by R01CA188506, R01CA176012, and a UCSD-Roche Extending Innovation Network (EIN) program grant (414951) to G.-S.F.

Received: November 18, 2015 Revised: October 20, 2016 Accepted: November 15, 2016 Published: December 13, 2016

REFERENCES

Aharoni-Simon, M., Hann-Obercyger, M., Pen, S., Madar, Z., and Tirosh, O. (2011). Fatty liver is associated with impaired activity of PPARg-coactivator 1 a (PGC1 a) and mitochondrial biogenesis in mice. Lab. Invest. 91,1018-1028.

Bard-Chapeau, E.A., Yuan, J., Droin, N., Long, S., Zhang, E.E., Nguyen, T.V., and Feng, G.S. (2006). Concerted functions of Gab1 and Shp2 in liver regeneration and hepatoprotection. Mol. Cell. Biol. 26, 4664-4674. Bard-Chapeau, E.A., Li, S., Ding, J., Zhang, S.S., Zhu, H.H., Princen, F., Fang, D.D., Han, T., Bailly-Maitre, B., Poli, V., et al. (2011). Ptpn11/Shp2 acts as a tumor suppressor in hepatocellular carcinogenesis. Cancer Cell 19, 629-639.

Brown, P.H., Alani, R., Preis, L.H., Szabo, E., and Birrer, M.J. (1993). Suppression of oncogene-induced transformation by a deletion mutant of c-jun. Oncogene 8, 877-886.

Cantley, L.C. (2002). The phosphoinositide 3-kinase pathway. Science 296, 1655-1657.

Chalasani, N., Younossi, Z., Lavine, J.E., Diehl, A.M., Brunt, E.M., Cusi, K., Charlton, M., and Sanyal, A.J. (2012). The diagnosis and management of non-alcoholic fatty liver disease: practice Guideline by the American Association for the Study of Liver Diseases, American College of Gastroenterology, and the American Gastroenterological Association. Hepatology 55, 20052023.

Chalhoub, N., and Baker, S.J. (2009). PTEN and the PI3-kinase pathway in cancer. Annu. Rev. Pathol. 4, 127-150.

Chan, R.J., and Feng, G.S. (2007). PTPN11 is the first identified proto-onco-gene that encodes a tyrosine phosphatase. Blood 109, 862-867. Chen, Y.N., LaMarche, M.J., Chan, H.M., Fekkes, P., Garcia-Fortanet, J., Acker, M.G., Antonakos, B., Chen, C.H., Chen, Z., Cooke, V.G., et al. (2016). Allosteric inhibition of SHP2 phosphatase inhibits cancers driven by receptor tyrosine kinases. Nature 535, 148-152.

Cross, D.A., Alessi, D.R., Cohen, P., Andjelkovich, M., and Hemmings, B.A. (1995). Inhibition of glycogen synthase kinase-3 by insulin mediated by protein kinase B. Nature 378, 785-789.

Eferl, R., and Wagner, E.F. (2003). AP-1: a double-edged sword in tumorigen-esis. Nat. Rev. Cancer 3, 859-868.

Feng, G.S. (2012). Conflicting roles of molecules in hepatocarcinogenesis: paradigm or paradox. Cancer Cell 21, 150-154.

Galicia, V.A., He, L., Dang, H., Kanel, G., Vendryes, C., French, B.A., Zeng, N., Bayan, J.A., Ding, W., Wang, K.S., et al. (2010). Expansion of hepatic tumor progenitor cells in Pten-null mice requires liver injury and is reversed by loss of AKT2. Gastroenterology 139, 2170-2182.

Gotschel, F., Kern, C., Lang, S., Sparna, T., Markmann, C., Schwager, J., McNelly, S., von Weizsacker, F., Laufer, S., Hecht, A., and Merfort, I. (2008). Inhibition of GSK3 differentially modulates NF-kappaB, CREB, AP-1 and beta-catenin signaling in hepatocytes, but fails to promote TNF-alpha-induced apoptosis. Exp. Cell Res. 314, 1351-1366.

Han, T., Xiang, D.M., Sun, W., Liu, N., Sun, H.L., Wen, W., Shen, W.F., Wang, R.Y., Chen, C., Wang, X., et al. (2015). PTPN11/Shp2 overexpression enhances liver cancer progression and predicts poor prognosis of patients. J Hepatol. 63, 651-660.

He, G., Yu, G.Y., Temkin, V., Ogata, H., Kuntzen, C., Sakurai, T., Sieghart, W., Peck-Radosavljevic, M., Leffert, H.L., and Karin, M. (2010). Hepatocyte IKKbeta/NF-kappaB inhibits tumor promotion and progression by preventing oxidative stress-driven STAT3 activation. Cancer Cell 17, 286-297.

He, G., Dhar, D., Nakagawa, H., Font-Burgada, J., Ogata, H., Jiang, Y., Shala-pour, S., Seki, E., Yost, S.E., Jepsen, K., et al. (2013). Identification of liver cancer progenitors whose malignant progression depends on autocrine IL-6 signaling. Cell 155, 384-396.

He, L., Gubbins, J., Peng, Z., Medina, V., Fei, F., Asahina, K., Wang, J., Kahn, M., Rountree, C.B., and Stiles, B.L. (2016). Activation of hepatic stellate cell in Pten null liver injury model. Fibrogenesis Tissue Repair 9, 8. Horie, Y., Suzuki, A., Kataoka, E., Sasaki, T., Hamada, K., Sasaki, J., Mizuno, K., Hasegawa, G., Kishimoto, H., lizuka, M., et al. (2004). Hepatocyte-specific Pten deficiency results in steatohepatitis and hepatocellular carcinomas. J. Clin. Invest. 113, 1774-1783.

Jiang, C., Hu, F., Tai, Y., Du, J., Mao, B., Yuan, Z., Wang, Y., and Wei, L. (2012). The tumor suppressor role of Src homology phosphotyrosine phosphatase 2 in hepatocellular carcinoma. J. Cancer Res. Clin. Oncol. 138, 637-646. Kaposi-Novak, P., Lee, J.S., Gomez-Quiroz, L., Coulouarn, C., Factor, V.M., and Thorgeirsson, S.S. (2006). Met-regulated expression signature defines a subset of human hepatocellular carcinomas with poor prognosis and aggressive phenotype. J. Clin. Invest. 116, 1582-1595.

Kenerson, H.L., Yeh, M.M., Kazami, M., Jiang, X., Riehle, K.J., McIntyre, R.L., Park, J.O., Kwon, S., Campbell, J.S., and Yeung, R.S. (2013). Akt and mTORC1 have different roles during liver tumorigenesis in mice. Gastroenter-ology 144, 1055-1065.

Lade, A., Noon, L.A., and Friedman, S.L. (2014). Contributions of metabolic dysregulation and inflammation to nonalcoholic steatohepatitis, hepatic fibrosis, and cancer. Curr. Opin. Oncol. 26, 100-107. Lai, L.A., Zhao, C., Zhang, E.E., and Feng, G.S. (2004). The Shp-2 tyrosine phosphatase. In Protein Phosphatases, J. Arino and D. Alexander, eds. (Springer), pp. 275-299.

Lanaya, H., Natarajan, A., Komposch, K., Li, L., Amberg, N., Chen, L., Wculek, S.K., Hammer, M., Zenz, R., Peck-Radosavljevic, M., et al. (2014). EGFR has a tumour-promoting role in liver macrophages during hepatocellular carcinoma formation. Nat. Cell Biol. 16, 972-981.

Lesche, R., Groszer, M., Gao, J., Wang, Y., Messing, A., Sun, H., Liu, X., and Wu, H. (2002). Cre/loxP-mediated inactivation of the murine Pten tumor suppressor gene. Genesis 32, 148-149.

Li, J., Chen, H., Tang, M.S., Shi, X., Amin, S., Desai, D., Costa, M., and Huang, C. (2004). PI-3K and Akt are mediators of AP-1 induction by 5-MCDE in mouse epidermal Cl41 cells. J. Cell Biol. 165, 77-86.

Li, S., Hsu, D.D., Li, B., Luo, X., Alderson, N., Qiao, L., Ma, L., Zhu, H.H., He, Z., Suino-Powell, K., et al. (2014). Cytoplasmic tyrosine phosphatase Shp2 coordinates hepatic regulation of bile acid and FGF15/19 signaling to repress bile acid synthesis. Cell Metab. 20, 320-332.

Liao, R., Sun, T.W., Yi, Y., Wu, H., Li, Y.W., Wang, J.X., Zhou, J., Shi, Y.H., Cheng, Y.F., Qiu, S.J., and Fan, J. (2012). Expression of TREM-1 in hepatic stellate cells and prognostic value in hepatitis B-related hepatocellular carcinoma. Cancer Sci. 103, 984-992.

Llovet, J.M., Ricci, S., Mazzaferro, V., Hilgard, P., Gane, E., Blanc, J.F., de Oli-veira, A.C., Santoro, A., Raoul, J.L., Forner, A., et al.; SHARP Investigators Study Group (2008). Sorafenib in advanced hepatocellular carcinoma. N. Engl. J. Med. 359, 378-390.

Maeda, S., Kamata, H., Luo, J.L., Leffert, H., and Karin, M. (2005). IKKbeta couples hepatocyte death to cytokine-driven compensatory proliferation that promotes chemical hepatocarcinogenesis. Cell 121, 977-990.

Michelotti, G.A., Machado, M.V., and Diehl, A.M. (2013). NAFLD, NASH and liver cancer. Nat. Rev. Gastroenterol. Hepatol. 10, 656-665. Min, L., Ji, Y., Bakiri, L., Qiu, Z., Cen, J., Chen, X., Chen, L., Scheuch, H., Zheng, H., Qin, L., et al. (2012). Liver cancer initiation is controlled by AP-1 through SIRT6-dependent inhibition of survivin. Nat. Cell Biol. 14,1203-1211. Nakagawa, H., Umemura, A., Taniguchi, K., Font-Burgada, J., Dhar, D., Ogata, H., Zhong, Z., Valasek, M.A., Seki, E., Hidalgo, J., et al. (2014). ER stress cooperates with hypernutrition to trigger TNF-dependent spontaneous HCC development. Cancer Cell 26, 331-343.

Neel, B.G., Gu, H., and Pao, L. (2003). The "Shp"ing news: SH2 domain-containing tyrosine phosphatases in cell signaling. Trends Biochem. Sci. 28, 284-293.

Park, E.J., Lee, J.H., Yu, G.Y., He, G., Ali, S.R., Holzer, R.G., Osterreicher, C.H., Takahashi, H., and Karin, M. (2010). Dietary and genetic obesity promote liver inflammation and tumorigenesis by enhancing IL-6 and TNF expression. Cell 140, 197-208.

Peloponese, J.M., Jr., and Jeang, K.T. (2006). Role for Akt/protein kinase B and activator protein-1 in cellular proliferation induced by the human T-cell leukemia virus type 1 tax oncoprotein. J. Biol. Chem. 281, 8927-8938. Pilati, C., Letouze, E., Nault, J.C., Imbeaud, S., Boulai, A., Calderaro, J., Poussin, K., Franconi, A., Couchy, G., Morcrette, G., et al. (2014). Genomic profiling of hepatocellular adenomas reveals recurrent FRK-activating mutations and the mechanisms of malignant transformation. Cancer Cell 25, 428-441. Ransone, L.J., and Verma, I.M. (1990). Nuclear proto-oncogenes fos and jun. Annu. Rev. Cell Biol. 6, 539-557.

Sanchez-Pareja, A., Clement, S., Peyrou, M., Spahr, L., Negro, F., Rubbia-Brandt, L., and Foti, M. (2016). Phosphatase and tensin homolog is a differential diagnostic marker between nonalcoholic and alcoholic fatty liver disease. World J. Gastroenterol. 22, 3735-3745.

Shi, Z.Q., Lu, W., and Feng, G.S. (1998). The Shp-2 tyrosine phosphatase has opposite effects in mediating the activation of extracellular signal-regulated and c-Jun NH2-terminal mitogen-activated protein kinases. J. Biol. Chem. 273, 4904-4908.

Shimizu, Y., Kinoshita, I., Kikuchi, J., Yamazaki, K., Nishimura, M., Birrer, M.J., and Dosaka-Akita, H. (2008). Growth inhibition of non-small cell lung cancer cells by AP-1 blockade using a cJun dominant-negative mutant. Br. J. Cancer 98, 915-922.

Song, M.S., Salmena, L., and Pandolfi, P.P. (2012). The functions and regulation of the PTEN tumour suppressor. Nat. Rev. Mol. Cell Biol. 13, 283-296. Stiles, B., Wang, Y., Stahl, A., Bassilian, S., Lee, W.P., Kim, Y.J., Sherwin, R., Devaskar, S., Lesche, R., Magnuson, M.A., and Wu, H. (2004). Liver-specific deletion of negative regulator Pten results in fatty liver and insulin hypersensitivity [corrected]. Proc. Natl. Acad. Sci. USA 101, 2082-2087. Takami, T., Kaposi-Novak, P., Uchida, K., Gomez-Quiroz, L.E., Conner, E.A., Factor, V.M., and Thorgeirsson, S.S. (2007). Loss of hepatocyte growth

factor/c-Met signaling pathway accelerates early stages of N-nitrosodiethyl-amine induced hepatocarcinogenesis. Cancer Res. 67, 9844-9851.

Theise, N.D. (2014). World Cancer Report 2014 IARC.

Wang, Q., Yu, W.N., Chen, X., Peng, X.D., Jeon, S.M., Birnbaum, M.J., Guzman, G., and Hay, N. (2016). Spontaneous hepatocellular carcinoma after the combined deletion of Akt isoforms. Cancer Cell 29, 523-535.

Wolf, M.J., Adili, A., Piotrowitz, K., Abdullah, Z., Boege, Y., Stemmer, K., Ringelhan, M., Simonavicius, N., Egger, M., Wohlleber, D., et al. (2014). Metabolic activation of intrahepatic CD8+T cells and NKTcells causes nonalcoholic steatohepatitis and liver cancer via cross-talk with hepatocytes. Cancer Cell 26, 549-564.

Worby, C.A., and Dixon, J.E. (2014). Pten. Annu. Rev. Biochem. 83, 641-669.

Yoshida, L.S., Miyazawa, T., Hatayama, I., Sato, K., Fujimoto, K., and Kaneda, T. (1993). Phosphatidylcholine peroxidation and liver cancer in mice fed a choline-deficient diet with ethionine. Free Radic. Biol. Med. 14, 191-199.

Zhang, E.E., Chapeau, E., Hagihara, K., and Feng, G.S. (2004). Neuronal Shp2 tyrosine phosphatase controls energy balance and metabolism. Proc. Natl. Acad. Sci. USA 101, 16064-16069.

Zhang, X.F., Tan, X., Zeng, G., Misse, A., Singh, S., Kim, Y., Klaunig, J.E., and Monga, S.P. (2010). Conditional beta-catenin loss in mice promotes chemical hepatocarcinogenesis: role of oxidative stress and platelet-derived growth factor receptor alpha/phosphoinositide 3-kinase signaling. Hepatology 52, 954-965.

Zhu, X., Jiang, S., Hu, N., Luo, F., Dong, H., Kang, Y.M., Jones, K.R., Zou, Y., Xiong, L., and Ren, J. (2014). Tumour necrosis factor-a inhibition with lenalido-mide alleviates tissue oxidative injury and apoptosis in ob/ob obese mice. Clin. Exp. Pharmacol. Physiol. 41, 489-501.

Zhu, H.H., Luo, X., Zhang, K., Cui, J., Zhao, H., Ji, Z., Zhou, Z., Yao, J., Zeng, L., Ji, K., et al. (2015). Shp2 and Pten have antagonistic roles in myeloprolifer-ation but cooperate to promote erythropoiesis in mammals. Proc. Natl. Acad. Sci. USA 112, 13342-13347.

Zucman-Rossi, J., Villanueva, A., Nault, J.C., and Llovet, J.M. (2015). Genetic landscape and biomarkers of hepatocellular carcinoma. Gastroenterology 149, 1226-1239.e4.