Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms

Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms

Article Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms Graphical Abstract Authors Qi Wang, Wan-Ni Yu, Xinyu Chen, ...

4MB Sizes 0 Downloads 68 Views

Article

Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms Graphical Abstract

Authors Qi Wang, Wan-Ni Yu, Xinyu Chen, ..., Morris J. Birnbaum, Grace Guzman, Nissim Hay

Correspondence [email protected]

In Brief Wang et al. show that systemic deletion of Akt1 and Akt2 is lethal while hepatic deletion leads to hepatocellular carcinoma (HCC) with complete penetrance in a FoxO1-dependent manner. Deletion of either Akt1 or Akt2 did not inhibit diethylnitrosamineinduced HCC, but Akt2 loss increased metastasis.

Highlights d

Early onset of HCC after hepatic deletion of Akt1 in Akt2 mice

Accession Numbers /

d

The HCC is preceded by liver injury and inflammation in a FoxO1-dependent manner

d

Akt2 deficiency promotes lung metastasis after hepatocarcinogenesis

d

Akt is required for hepatic STAT5 phosphorylation and IGF1 expression

Wang et al., 2016, Cancer Cell 29, 1–13 April 11, 2016 ª2016 Elsevier Inc. http://dx.doi.org/10.1016/j.ccell.2016.02.008

GSE77862

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

Cancer Cell

Article Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms Qi Wang,1,5 Wan-Ni Yu,1,5 Xinyu Chen,1 Xiao-ding Peng,1 Sang-Min Jeon,1,6 Morris J. Birnbaum,3 Grace Guzman,4 and Nissim Hay1,2,* 1Department

of Biochemistry and Molecular Genetics, College of Medicine, University of Illinois at Chicago, Chicago, IL 60607, USA & Development Section, Jesse Brown VA Medical Center, Chicago, IL 60612, USA 3Division of Endocrinology, Diabetes and Metabolism, Department of Medicine, Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA 19104, USA 4Department of Pathology, College of Medicine, Cancer Center, University of Illinois Hospital and Health Science Chicago, Chicago, IL 60612, USA 5Co-first author 6Present address: College of Pharmacy, Ajou University Yeongtong-gu, Suwon-si, Gyeonggi-do 443-749, Korea *Correspondence: [email protected] http://dx.doi.org/10.1016/j.ccell.2016.02.008 2Research

SUMMARY

Akt is frequently hyperactivated in human cancers and is targeted for cancer therapy. However, the physiological consequences of systemic Akt isoform inhibition were not fully explored. We showed that while combined Akt1 and Akt3 deletion in adult mice is tolerated, combined Akt1 and Akt2 deletion induced rapid mortality. Akt2 / mice survived hepatic Akt1 deletion but all developed spontaneous hepatocellular carcinoma (HCC), which is associated with FoxO-dependent liver injury and inflammation. The gene expression signature of HCC-bearing livers is similar to aggressive human HCC. Consistently, neither Akt1 / nor Akt2 / mice are resistant to diethylnitrosamine-induced hepatocarcinogenesis, and Akt2 / mice display a high incidence of lung metastasis. Thus, in contrast to other cancers, hepatic Akt inhibition induces liver injury that could promote HCC.

INTRODUCTION The serine/threonine kinase, Akt, is perhaps the most frequently activated oncoprotein in human cancers. Akt is hyperactivated by multiple mechanisms, such as by mutations that hyperactivate the catalytic subunits of phosphatidylinositol 3-kinase (PI3K), mutations that inactivate the tumor suppressor PTEN, activation of growth factor receptors, and activating mutations in Akt isoforms themselves. The hyperactivation of PI3K/Akt signaling and its downstream effector mTORC1 in human cancers have prompted approaches that target this pathway for cancer therapy. Currently there are multiple clinical trials aimed at assessing the efficacy of PI3K/Akt inhibitors in cancer patients

(Dienstmann et al., 2014; Fruman and Rommel, 2014). However, pan-PI3K and pan-Akt inhibitors inhibit the activity of the three Akt isoforms. Therefore, it became necessary to understand the consequences of systemic Akt isoform deletion on the survival and physiology of adult mice. The three Akt isoforms (Akt1–3), which are encoded by separate genes, share greater than 80% identity at the amino acid level. Gene targeting of the individual Akt isoforms in mice resulted in distinct phenotypes. Therefore, it is important to determine whether systemic Akt inhibition is therapeutic without eliciting adverse physiological consequences. Genetic studies in mice have demonstrated that the Akt1 deletion is well tolerated and can inhibit development of cancer in various mouse models

Significance The results presented here suggest that the development of Akt-isoform-specific inhibitors should be considered. The liver injury, inflammation, and unexpected spontaneous HCC that develop when both hepatic Akt1 and Akt2 are deleted might have implications for cancer therapy. Moreover, while neither Akt1 nor Akt2 are required for hepatocarcinogenesis, the deficiency of Akt2 promotes a high incidence of lung metastasis following hepatocarcinogenesis. The results raise the possibility that systemic administration of pan-PI3K/Akt inhibitors might increase the risk for liver injury and inflammation, and possibly HCC, if hepatic Akt activity is markedly decreased. The results suggest that hepatic Akt inhibition in fatty liver could be one mechanism by which obesity promotes HCC and that pan-PI3K/Akt inhibitors could have undesired outcomes especially for obesity-mediated HCC. Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc. 1

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

(Chen et al., 2006; Hollander et al., 2011; Ju et al., 2007; Maroulakou et al., 2007; Skeen et al., 2006). By contrast, Akt2 deletion elicits insulin resistance and does not inhibit cancer development in mouse models (Maroulakou et al., 2007; Xu et al., 2012). However, those studies utilized germline deletion of Akt isoforms, and thus they do not address the effect of Akt isoforms on tumor progression and do not emulate the drug therapy that is administered after tumor detection. In addition, germline deletion of both Akt1 and Akt2 is neonatal lethal, and the deletion of both Akt1 and Akt3 is embryonic lethal, while the deletion of both Akt2 and Akt3 does not cause lethality (reviewed in Hay, 2011). However, the consequences of systemic combined deletions of Akt isoforms in adult mice, which are more relevant to cancer therapy, are not known. To address these issues, we deleted Akt isoforms systemically in adult mice. We recently showed that systemic deletion of Akt1 in adult p53 / mice after tumor onset halted and regressed thymic lymphoma developed in these mice (Yu et al., 2015). In the present studies, we determined the consequences of combined Akt isoforms deletion in adult mice. RESULTS Combined Deletion of Akt1 and Akt2 in Adult Mice Induces Rapid Mortality We addressed the effects of individual Akt isoform or combined Akt isoform systemic deficiency in adult mice. We generated Akt1f/f;R26CreERT2 mice, in which a construct expressing Cre recombinase fused with a mutated estrogen receptor ligand-binding domain that can be activated by tamoxifen (CreERT2) was knocked into the Rosa26 (R26) locus. Thus, upon injection of tamoxifen, Cre is activated to systemically delete the floxed allele (Ventura et al., 2007). We have shown that Akt1 can be systemically deleted in these adult mice following injection of tamoxifen, and that the deletion did not elicit any overt phenotype but was able to regress thymic lymphoma in Trp53 / mice (Yu et al., 2015). Akt1f/f;R26CreERT2 mice were crossed with either Akt2 / or Akt3 / mice to generate either Akt1f/f; Akt2 / ;R26CreERT2 or Akt1f/f;Akt3 / ;R26CreERT2 mice (Figure 1A). The mice received tamoxifen at 2 months of age. Surprisingly, the Akt3 / mice survived the deletion of Akt1 without any overt phenotype (Figure 1B). By contrast, the Akt2 / mice did not survive the Akt1 systemic deletion and died within 2 weeks or 2 months after tamoxifen injection, depending on the level of Cre recombinase expression (homozygous versus heterozygous; Figure 1B). The mortality appeared to be independent of the age at which the deletion of Akt1 was induced because we found that 6- to 8-month-old mice also died with similar kinetics. Immediately after the induction of the systemic deletion of Akt1 in Akt2 / mice, we observed a transient increase in fed glucose levels that was followed by a decrease in glucose levels and hypoglycemia, depending on the level of Cre recombinase (Figure 1C). A marked decrease in body weight was observed after the deletion of Akt1 in Akt2 / mice (Figure 1D). To further assess the effect of the combined Akt1 and Akt2 deletion in adult mice, we generated Akt1f/f;Akt2f/f; R26CreERT2 mice in which both Akt1 and Akt2 could be systemically deleted. Tamoxifen injection in Akt1f/f;Akt2f/f;R26CreERT2 mice induced extremely rapid mortality (Figure 1E) and a rapid 2 Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc.

loss of body weight (Figure S1A). Notably, Akt1f/f;Akt2f/f; R26CreERT2 mice died faster than Akt1f/f;Akt2 / ;R26CreERT2 mice following tamoxifen administration. Thus, the rapid mortality was not due to the germline deletion of Akt2 and was also observed following acute systemic deletion of both Akt1 and Akt2. Interestingly, we observed a dramatic loss of fat tissue after systemic deletion of both Akt1 and Akt2 (Figure S1B), suggesting the utilization of fat for fatty acid oxidation and possibly a block in adipocyte differentiation, as observed in newborn mice carrying a germline deletion of both Akt1 and Akt2 (Peng et al., 2003; Nogueira et al., 2012). We measured the glucose and insulin levels in these mice and found that both were elevated immediately after tamoxifen injection and reduced to below normal levels before death (Figures 1F and 1G). We also observed increased circulating levels of interleukin-6 (IL-6), with dramatic elevations prior to death, suggesting severe inflammation possibly as a consequence of tissue damage (Figure 1H). Notably, systemic deletion of both Akt1 and Akt2 in adult mice was conducted in the FVB background, while the Akt1f/f;Akt2 / ;R26CreERT2 mice were in the C57BL/6 background, indicating that the observed phenotype was independent of the mouse strain. Finally, we found that the mortality rates of male and female mice were similar following systemic Akt1 and Akt2 deletion. We concluded that although both Akt1 and Akt3 are required for embryonic development, they are not required for adult viability. By contrast, germline deletion of both Akt1 and Akt2 is not required for the development of embryos to term but is required for adult viability, and thus, complete ablation of the activity of Akt1 and Akt2 is not tolerated in adult mice. The pan-Akt inhibitor MK2206 has been used in vivo in mice with doses ranging from 60 mg/kg to 300 mg/kg. One of the most commonly used doses is 120 mg/kg orally administered every 2 days. When we doubled this dose for a period of 3 weeks, MK2206 administration recapitulated the systemic deletion of Akt1 and Akt2 and reduced body weight (Figure S1C), and markedly induced IL-6 levels (Figure S1D). Although, we do not know the exact cause of death in response to the systemic deletion of Akt1 and Akt2, we hypothesize that the mice could not utilize external nutrients or absorb food; thus, they consumed fat as an energy and nutrient source. However, following the depletion of fat, they become hypoglycemic and die. Consistently, we found that the structure of the villi in the intestine was severely impaired following the induction of the Akt1 and Akt2 deletion (Figure S1E). This phenomenon was likely due to the inhibition of crypt cell proliferation, as measured by Ki67 staining (Figure S1F). Impaired intestinal structure and crypt cell proliferation was also observed after MK2206 administration (Figure S1G). Taken together, these results indicate that systemic inhibition of Akt1 and Akt2 could exert severe toxicity manifested by tissue damage, inflammation, and hypoglycemic shock. Hepatic Deletion of Akt1 in Akt2–/– Mice Caused Severe Growth Retardation As shown in Figure 1, systemic deletion of both Akt1 and Akt2 in adult mice induced rapid mortality. However, hepatic deletion of Akt1 by albumin promoter-driven Cre in Akt2 / mice (Akt1hep / ;Akt2 / ) was tolerated. The Akt1hep / ;Akt2 / mice had higher blood glucose levels than Akt2 / mice (Figure S2A) that were similar to those previously described for

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

A

E

B

F

C

G

D

Figure 1. The Combined Deletion of Akt1 and Akt2 in Adult Mice Induces Rapid Mortality Preceded by Body Weight Loss and Hypoglycemia (A) Breeding scheme to generate Akt1f/f;Akt2 / ; R26CreERT2 and Akt1f/f;Akt3 / ;R26CreERT2 mice. (B) Kaplan-Meier plot showing the survival of 2-month-old Akt1f/f;Akt2 / ;R26CreERT2+/ (n = 20), Akt1f/f;Akt2 / ;R26CreERT2+/+ (n = 10), and Akt1f/f; Akt3-/;-R26CreERT2+/+ (n = 10) mice after tamoxifen injection for 5 consecutive days. (C) Blood glucose levels in fed Akt1f/f;Akt2 / , Akt1f/f;Akt2 / R26CreERT2+/ , and Akt1f/f;Akt2 / ; R26CreERT2+/+ mice before and after tamoxifen injection (n = 6, ±SEM, p < 0.05 versus before tamoxifen). (D) Body weights of Akt1f/f;Akt2 / , Akt1f/f;Akt2 / ; R26CreERT2+/ , and Akt1f/f;Akt2 / R26CreERT2+/+ mice after tamoxifen injection (n = 6, ±SEM, p < 0.05 versus Akt1f/f;Akt2 / mice). (E) Kaplan-Meier plot showing the survival of 2-month-old Akt1f/fAkt2 / R26CreERT2+/+ mice after tamoxifen injection (n = 11). (F) Glucose levels in Akt1f/f;Akt2 / ;R26CreERT2+/+ mice before and after the start of tamoxifen injection (n = 4, ±SEM, p < 0.05 tamoxifen versus corn oil). (G) Serum insulin levels in Akt1f/f;Akt2 / ; R26CreERT2+/+ mice before and after the start of tamoxifen injection and before death (n = 4, ±SEM, p < 0.03 tamoxifen versus corn oil). (H) Serum IL-6 levels in tamoxifen-injected Akt1f/f; Akt2 / ;R26CreERT2+/+ mice (n = 4, ±SEM, p < 0.03 tamoxifen versus corn oil). See also Figure S1.

H

Akt2 / mice in which Akt1 was deleted by a-fetoprotein (Afp) promoter-driven Cre (Lu et al., 2012). The mice also displayed glucose intolerance (Figure S2B). Despite higher glucose levels, the high insulin level detected in Akt2 / mice was not significantly different after hepatic Akt1 deletion, and thus, Akt1hep / ; Akt2 / , and Akt2 / mice were phenotypically similar in terms of the insulin tolerance (Figures S2C and S2D). Akt1hep / Akt2 / mice displayed marked growth retardation with a decrease in body weight of approximately 40% compared with the mice with other genotypes (Figure S3A). The Akt1hep / ; Akt2 / mice had markedly reduced levels of IGF1 in the blood (Figure S3B), which likely resulted in the growth retardation. The liver is the major organ responsible for generating circulating IGF1, wherein expression is transcriptionally induced primarily by growth hormone (GH) (reviewed in Herrington et al., 2000; Klammt et al., 2008). However, circulating GH levels were not significantly different between Akt1hep / ;Akt2 / and Akt2 / or wild-type (WT) mice (Figure S3C). Nonetheless, Igf1 mRNA levels in the livers of Akt1hep / ;Akt2 / mice were markedly

lower than those in the livers of Akt2 / mice (Figure 2A). GH induces IGF1 expression by binding to its cognate receptor on hepatocytes and activating the STAT5 transcription factor, which in turn translocates into the nucleus to induce Igf1 transcription (Leung et al., 2000). We found that STAT5 tyrosine phosphorylation, which is readout for its activity, was severely inhibited in the livers of Akt1hep / ;Akt2 / mice (Figure 2B). Thus, in addition to the known effect of GH, Akt activity was also required for STAT5 activation and the expression in the liver. GH receptor (GHR) expression has been reported to be induced by insulin (Leung et al., 2000) and is a potential transcriptional target of FoxO1 in the liver, as determined by chromatin immunoprecipitation sequencing (ChIPseq) (Shin et al., 2012). Hepatic GHR has also been reported to be downregulated in a FoxO1-dependent manner after hepatic deletion of IRS1 and IRS2 (Dong et al., 2008). Because FoxO transcription factors are phosphorylated and inhibited by the insulin-Akt axis, we examined the expression of GHR in the livers of Akt1hep / ;Akt2 / mice. Indeed, we found that the mRNA level of GHR was significantly reduced in Akt1hep / ;Akt2 / mouse livers compared with Akt2 / mouse livers (Figure 2C). Thus, similar to the results reported for the hepatic deletion of IRS1 and IRS2 (Dong et al., 2008), GHR expression was suppressed in the absence of hepatic Akt1 and Akt2, likely by FoxO. Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc. 3

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

A

D

B

C

E

F

Figure 2. Hepatic Deletion of Akt1 in Akt2–/– Mice Inhibits Growth Hormone Receptor Expression, STAT5 Activation, and IGF1 Expression, and Induces Early Onset of HCC (A) qRT-PCR showing hepatic Igf1 mRNA levels in Akt2 / and Akt1hep / ;Akt2 / mice (n = 4, ±SEM, p = 0.0067). (B) Hepatic STAT5 phosphorylation on Tyr694 in Akt2 / or Akt1hep / ;Akt2 / mice. (C) Level of growth hormone receptor mRNA in Akt2 / and Akt1hep / ;Akt2 / mouse livers (n = 4, ±SEM, *p = 0.013). (D) Model illustrating the role of Akt in the GH/IGF1 axis (see text for details). (E) Representative images of the livers of 6-monthold male mice of the indicated genotype. (F) Quantification of the number of macroscopic tumors per liver in 6- to 7-month-old male mice (n = 10, ±SEM, **p < 0.0001 Akt1hep / ;Akt2 / mice versus Akt2 / ). (G) Liver sections showing tumor lesions in 4- and 7-month-old Akt1hep / Akt2 / mice (upper panel, low magnification; bottom panel, high magnification). The tumor areas are marked. See also Figures S2 and S3.

G

This phenomenon could explain, at least in part, the requirement for Akt for GH-mediated expression of IGF1 in the liver (Figure 2D). However, we cannot completely exclude the possibility of additional mechanisms by which Akt affects STAT5 phosphorylation (see Discussion). Hepatic Deletion of Akt1 in Akt2–/– Mice Induces Early Onset of Spontaneous HCC, Liver Injury, and Inflammation Further analyses of the Akt1hep / ;Akt2 / mice revealed that all the Akt1hep / ;Akt2 / male mice developed HCC by 6–7 months of age (8–9 months for females; Figure 2E). This finding was not observed in Akt2 / mice, even when one allele of Akt1 was deleted in the liver of the Akt2 / mice (Akt1hep+/-; Akt2 / mice) or when Akt1 was deleted in the liver of Akt2+/ mice (Akt1hep / ;Akt2+/ mice) (Figures 2E and S3D). An average of 19 macroscopic tumors per liver was detected in the Akt1hep / ;Akt2 / mice, and no macroscopic tumors were observed in the mice with other genotypes (Figure 2F). Importantly, the hepatocarcinogenesis observed in Akt1hep / ;Akt2 / 4 Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc.

mice was a relatively rapid process compared with most mouse models of hepatocarcinogenesis (Heindryckx et al., 2009), and microscopic neoplastic lesions had been observed in 3- to 4-month-old male mice (Figure 2G). By comparison, hepatic activation of Akt by hepatic deletion of Pten induces HCC after 75 weeks (Horie et al., 2004). In both the 4-monthold and 7-month-old mice, two major types of tumors, clear cell and trabecular, were observed (Figure 2G). Akt2 is the major Akt isoform expressed in the liver, and its deletion substantially reduced hepatic Akt activity, but this is not sufficient to induce HCC. Since even the deletion of Akt2 together with heterozygous deletion of Akt1 is not sufficient to induce HCC, we concluded that HCC is induced only when hepatic Akt activity is reduced below a threshold level. Considering that Akt is frequently hyperactivated in human cancers, these results were unexpected. The initiation of and predisposition to liver carcinogenesis is often associated with liver injury and inflammation. Consistently, we found that serum levels of the liver enzymes alanine transaminase (ALT) and aspartate transaminase (AST) were markedly elevated in Akt1hep / ;Akt2 / mice (Figures 3A and 3B), indicating liver damage, which was not observed in either Akt1hep+/-; Akt2 / or Akt2 / mice. Moreover, serum IL-6 levels (Figure 3C) and liver tumor necrosis factor alpha (TNF-a) mRNA (Figure 3D) were highly elevated in Akt1hep / ;Akt2 / mice compared with Akt2 / mice, suggesting inflammation. Consistent with the identified inflammation, F4/80 staining revealed numerous liver macrophages (Kupffer cells) in the tumor sections obtained from the livers of 3- to 4-month-old and 7- to 8-month-old Akt1hep-/;-Akt2 / mice (Figures 3E and 3F). In addition, plasma

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

A

B

C

AST (U/L)

ALT (U/L)

Serum IL-6 (pg/mL)

Akt2-/250

Akt1hep-/-;Akt2-/-

200 150 100 50 0

Akt1hep-/-; Akt2 -/-

F

Akt1hep-/- ; Akt2 -/-

Akt1hep+/-; Akt2 -/-

Akt2 -/ -

Akt2-/-

Akt1hep+/- ; Akt2 -/-

Akt2-/-

E

Akt1hep-/-;Akt2-/-

2.5 2.0

Akt1hep-/-;Akt2-/-

N

H&E

1.5

T

1.0 0.5

TNFα

Akt2-/-

F4/80

/

0.0

/

Relative fold change

D

Akt2 -/ -

Akt1hep-/-;Akt2-/-

N

N Akt2-/F4/80 positive (cells/HPF)

H&E

T N

F4/80

T G

Akt2-/-

800

Akt1hep-/-;Akt2-/-

*

600 400 200 0

Akt1hep-/-;Akt2-/40

*

N T

20

10

CC3

T

caspase3 positive (cells/HPF)

0

N

(A) Serum AST levels in 3- and 6-month-old male mice with the indicated genotypes (n = 4–6, ±SEM, p < 0.05 Akt1hep / ;Akt2 / versus Akt1hep+/-;Akt2 / or Akt2 / mice). (B) Serum ALT levels in 3- and 6-month-old male mice with the indicated genotypes (n = 4–6, ±SEM, p < 0.05 Akt1hep / ;Akt2 / versus Akt1hep+/-;Akt2 / or Akt2 / mice). (C) Serum IL-6 levels in 4- to 6-month-old Akt1hep / ;Akt2 / and Akt2 / male mice (n = 6, ±SEM, p < 0.05 Akt1hep / ;Akt2 / versus Akt2 / mice). (D) TNF-a mRNA levels as measured by qRT-PCR in the livers of 4- to 6-month-old Akt1hep / ;Akt2 / and Akt2 / male mice (n = 4, ±SEM, p = 0.045). (E) Representative images of H&E and F4/80 staining in liver sections from 3- to 4-month-old Akt1hep / ;Akt2 / and Akt2 / mice (N, non-tumor area; T, tumor area). Scale bar, 100 mm. (F) Images of H&E and F4/80 staining of liver sections from 7- to 8-month-old Akt1hep / ;Akt2 / and Akt2 / mice. Scale bar, 100 mm. Bar graph shows the quantification of F4/80 staining (n = 4, ±SEM, *p = 0.024). (G) Images of liver sections from 7-to 8-month-old Akt1hep / ;Akt2 / and Akt2 / male mice showing immunohistochemistry with either anti-Ki67 or anticleaved caspase 3 (CC3). Scale bar, 100 mm. Bar graphs shows quantifications of the results (n = 4, ±SEM; *p = 0.031 **p = 0.0005). See also Figure S4.

30

Ki67 positive (cells/HPF)

Ki67

Akt2-/Akt1hep-/-;Akt2-/-

T

Figure 3. Hepatic Deletion of Akt1 in Akt2–/– Mice Induces Liver Injury, High Serum Levels of IL-6, Inflammation, Cell Death, and Cell Proliferation

25 20 15 10 5 0

cells were frequently observed in the livers of Akt1hep / ;Akt2 / mice (Figure S4A). The tumors cells in the livers of Akt1hep / ; Akt2 / mice were highly proliferative, as indicated by Ki67 staining (Figure 3G). Interestingly, a high level of apoptosis, as measured by cleaved caspase 3, was observed mostly outside the tumor lesions in the livers of Akt1hep / ;Akt2 / mice, while Ki67 staining was observed mostly inside the tumors (Figures 3G and S4B). Taken together, these results indicate that the deletion of Akt1 in the livers of Akt2 / mice induces liver damage followed by inflammation and HCC development (Li et al., 2012). Liver Injury and Inflammation Induced by Hepatic Deletion of Akt1 and Akt2 Is FoxO-Dependent We conducted RNA-seq analyses of the livers of 6-month-old Akt2 / and Akt1hep / ;Akt2 / mice. Although Akt2 was the major isoform expressed in the liver, the deletion of Akt1 in the livers of Akt2 / mice significantly and markedly induced many FoxO

target genes (Figure S5). We confirmed the induction of several FoxO target genes, such as Rbp1, Irs2, Igfbp1, and Sgk1, by qRT-PCR (Figure 4A). We also found that the pro-apoptotic FoxO target genes Fasl and Bcl2l11 (Bim) were significantly versus induced in Akt1hep / ;Akt2 / Akt2 / livers (Figure 4B), raising the possibility that liver injury and inflammation induced by hepatic deletion of Akt1 and Akt2 were dependent on FoxO. To verify this possibility, we employed Akt1f/f;Akt2f/f and Akt1f/f;Akt2f/f;Foxo1f/f mice (Lu et al., 2012). One-month-old mice were systemically injected with AAV8-Tbg-Cre, which is an adeno-associated virus that expresses Cre driven by the hepatocyte-specific promoter of the thyroxine-binding globulin (Tbg) gene. AAV8-Tbg-GFP was used as a control (Figure 4C). This strategy has been used previously to efficiently delete both Akt1 and Akt2 in hepatocytes (Lu et al., 2012). Hepatic deletion of both Akt1 and Akt2 induced liver injury, as manifested by high levels of ALT, which was not observed after combined hepatic deletion of Akt1, Akt2, and Foxo1 (Figure 4D). Consequently, hepatic deletion of both Akt1 and Akt2 also induced liver inflammation, which was not observed in the absence of FoxO1 (Figures 4E and 4F). The results indicate that even in adult mice the hepatic deletion of Akt1 and Akt2 elicits liver injury and inflammation, which are

**

Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc. 5

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

Akt2-/Akt1hep-/-;Akt2-/-

**

4

2

**

4

2

0

0

B 2.0

Fasl

Bcl2l11

1.5

1.0

0.5

0.0

140 120 100 80 2 .0

Igfbp1

*

1 .5 1 .0 0 .5 0 .0

1 .5

* Relative fold change

Relative fold change

Irs2 6

Relative fold change

Relative fold change

Rbp1

Relative fold change

6

8

Relative fold change

A

*

20

Sgk1

15

10

5

C AAVLiverCRE

AAVLiverCRE

1 .0

0 .5

Akt1f/f;Akt2f/f

Akt1f/f;Akt2f/f;Foxo1f/f

150

E

* ALT (U/L)

*

0

0 .0

D

Figure 4. Liver Injury and Inflammation Induced by Hepatic Deletion of Akt1 and Akt2 Are FoxO1-dependent

100

Akt1f/f;Akt2f/f +Cre

(A) The mRNA level of known FoxO target genes in the livers of Akt1hep / ;Akt2 / compared with Akt2 / mice was measured by qRT-PCR (n = 4 ±SEM, *p < 0.05, **p < 0.01). (B) Fasl and Bcl2l11 mRNAs in the livers of Akt1hep / ;Akt2 / and Akt2 / mice were measured by qRT-PCR (n = 4, ±SEM, *p < 0.05). (C) Schematic depicting systemic injection of AAVTbg-Cre or control AAV-Tbg-GFP for the hepatic deletions of either both Akt1 and Akt2 or Akt1, Akt2, and Foxo1 in Akt1f/f;Akt2f/f and Akt1f/f;Akt2f/f; Foxo1f/f mice, respectively. (D) Serum ALT levels after AAV-Tbg-Cre or AAVTbg-GFP injection (n = 4, ±SEM, *p < 0.05). (E) Representative liver section images immunostained with anti-F4/80 after deletion of either both Akt1 and Akt2 or Akt1, Akt2, and FoxO1. Scale bar, 100 mm. (F) Quantification of F4/80 immunostaining (n = 4, ±SEM, **p < 0.005, GFP and Akt1f/f;Akt2f/f versus Akt1f/f;Akt2f/f;Foxo1f/f mice). See also Figures S5 and S6.

50

0

cells via STAT3 activation (reviewed in Taniguchi and Karin, 2014). ConsisAkt1f/f;Akt2f/f Akt1f/f;Akt2f/f;Foxo1f/f tently, we detected relatively high levels f/f;Akt2f/f;Foxo1f/f +Cre of STAT3 phosphorylation in the livers of Akt1 F Akt1hep / ;Akt2 / mice and, in general, 100 higher STAT3 phosphorylation in tumors ** 80 compared with non-tumor areas (Fig60 ure 5A). We found that apoptosis occurred 40 mostly outside the area of the tumor and 20 proliferation occurred mostly inside the tumor (Figures 3G and S4B). Therefore, 0 GFP Cre GFP Cre we speculated that the dead cells could Akt1f/f;Akt2f/f Akt1f/f;Akt2f/f;Foxo1f/f induce inflammation and higher levels of IL-6, which in turn activated STAT3 in the pro-tumorigenic. Indeed, these mice eventually develop macro- survived cells that then give rise to tumors. Although we initially scopic tumors with an average of two to three tumors per liver observed relatively high Akt1 expression in the tumor-bearing (Figure S6). While AAV8-Tbg-Cre-injected Akt1f/f;Akt2f/f mice livers (Figure S7A), this expression originated from non-hepatoeventually develop liver tumors, Akt1f/f;Akt2f/f;Foxo1f/f mice did cytes such as hematopoietic cells, fibroblasts, and endothelial not, although some abnormalities were observed (Figure S6A). cells, which reside at higher levels in the tumor area. When liver This is also manifested by the dramatic increase in Ki67 staining tumor sections were subjected to immune-histo-cytochemistry in the Akt1hep / ;Akt2hep / tumor-bearing livers in comparison with Akt1 antibodies, only the non-hepatocyte cells within the with Akt1hep / ;Akt2hep / ;Foxo1hep / livers (Figure S6B). tumor area were stained positive (Figure 5B). Co-immunostainConsistently, Akt1hep / ;Akt2hep / mice display high levels of ing with F4/80 and Akt1 clearly showed that the Akt1 positive circulating IL-6, while Akt1hep / ;Akt2hep-/;-Foxo1hep / mice cells are macrophages (Figure S7B). When hepatocytes derived did not (Figure S6C). Taken together, these results show that from the tumors were cultured, we found almost no Akt1 expreshepatic deletion of both Akt1 and Akt2 induces chronic liver sion and Akt phosphorylation in comparison with hepatocytes injury, inflammation, and HCC, which are largely dependent on derived from Akt2 / liver (Figures 5C, S7C, and S7D). The residual Akt phosphorylation that was observed after cell culturing in the activation of FoxO1. some of the samples cannot be derived from Akt3 since both normal hepatocytes and HCC cells do not expresses detectable Liver Tumors that Develop in Akt1hep–/–Akt2–/– Mice level of Akt3 (Figure S7D). It is therefore likely that the observed Display Relatively High Levels of STAT3 Phosphorylation minimal Akt1 expression is derived from a small population of and the Expression of Genes Commonly Induced in HCC A high level of IL-6 in response to liver damage is pro-onco- non-hepatocyte cells that were co-cultured with the hepatogenic in the liver, and it can induce the proliferation of liver cytes. Genotyping analysis of the cultured tumor hepatocytes Cre

GFP

Cre

F4/80 positive (cells/HPF)

GFP

6 Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc.

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

Akt1hep+/-;Akt2-/-

A

T

NT

T

NT

Figure 5. Phospho-STAT3, and IL-6 Expression in Akt1hep–/–;Akt2–/– Liver

Akt2-/-

Akt1hep-/-;Akt2-/T

p-STAT3 STAT3

D

500 µm

Akt1

Floxed NS

Akt2

Deleted pAkt 100 µm

pSTAT3

Akt1 hep -/;Akt2 -/-

Akt1 f/f ;Akt2 -/-

C

Akt1 f/f ;Akt2 -/-

B

Akt1 hep -/;Akt2 -/-

actin

(A) Immunoblot showing the level of STAT3 phosphorylation in the livers of 6-month-old male mice with the indicated genotypes (NT, non-tumor area; T, tumor area). (B) Representative liver tumor section showing immunohistochemistry with Akt1 antibodies. (C) Immunoblot showing the expression and phosphorylation of Akt1, Akt2, and STAT3 in isolated hepatocytes derived from Akt1f/f;Akt2 / liver and isolated hepatocytes derived from tumors in Akt1hep / ;Akt2 / liver. (D) PCR genotyping showing the floxed and deleted Akt1 in hepatocytes isolated from Akt1f/f;Akt2 / liver and from tumors in Akt1hep / ;Akt2 / liver. NS, non-specific band. (E) Representative images of IHC for IL-6 of Akt1hep / ;Akt2 / liver sections from different liver samples. Scale bar, 100 mm. (F) A representative image showing positively stained IL-6 cells and the location of the tumor (marked with a dashed line). Scale bar, 500 mm. Inset shows a higher magnification of the IL-6positive cells. Scale bar, 100 mm. See also Figure S7.

E STAT3 Actin

F

is consistent with this notion (Figures 5D and S7E). This is consistent with previously described hepatic deletion of genes that promote liver cancer and in which residual expression of the deleted genes is due to the present of non-hepatocyte cells (Luedde et al., 2007; Umemura et al., 2014). We therefore concluded that the hepatic tumor cells survive and proliferate despite Akt ablation, possibly through other mutations and/or STAT3 activation. Interestingly, we found that IL-6 was also produced by hepatocytes in the liver, indicating autocrine IL-6 production (Figure 5E), although these hepatocytes were usually located in areas distant from the tumor area (Figure 5F). Analysis of the RNA-seq data revealed that the expression of many genes that are induced in diethylnitrosamine (DEN)mediated HCC and in human HCC (He et al., 2013) are also induced in the livers of Akt1hep / ;Akt2 / mice. For instance, genes that are induced in DEN-mediated HCC, such as trefoil factor 3 (Tff3), the serine protease inhibitor Spink1, Ly6d, glypi-

can 3 (Gpc3), alpha-fetoprotein (Afp), and Cd44, are also induced in the livers of Akt1hep / ;Akt2 / mice (Figure 6A). Thorgeirsson and colleagues have analyzed gene expression in human HCC and based on this analysis, they found a distinct subtype of aggressive HCC (Lee et al., 2006). This subtype, which is associated with poor prognosis, expresses several embryonic genes and is characterized by the regulation of a network of genes that are associated with the induction of Fos, and FosL2 expression. Interestingly, we found that Fos and Fosl2 expression, as well as expression of several other genes in the network that are considered Fos and FosL2 or AP-1 targets, were markedly induced in Akt1hep / ;Akt2 / livers bearing tumors (Figures S5 and 6B). Finally, in addition to Afp, H19, and Gpc3, we discovered several other fetal genes that were induced in human HCC, such as Scd2 and Bex1 (Figure S8). An increase in Igf2bp1 expression was observed (Figure 6C), but the most profound finding was the induction of Igf2bp3 mRNA and protein in Akt1hep / Akt2 / livers (Figures 6D and 6E). The IGF2BPs, which include IGF2BP1, IGF2BP2, and IGF2BP3, were originally discovered as RNA binding proteins, which stabilize IGF2 mRNA. However, they also affect mRNA stability and the expression of many other genes (reviewed in Bell et al., 2013; Lederer et al., 2014). While IGF2BP2 is widely expressed in all mouse adult tissues, IGF2BP1 and IGF2BP3 are expressed mostly in embryonic tissues and are not expressed in normal adult liver (Lederer et al., 2014). However, a high expression level of IGF2BP1 and IGF2BP3 has been observed in many human Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc. 7

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

Figure 6. Characterization of Akt1hep–/–; Akt2–/– Liver-Bearing Tumors by Gene Expression

Akt2-/Akt1hep-/-;Akt2-/-

**

100 80 5 4 3 2 1

200 150 100 5 4 3 2 1

Relative fold change

15

*

10

*

1000

100

50

Cd44

5

*

5

0

15

*

Relative fold change

15

Fosl2

Nr4a1

10

5

*

900 800 700 600 500 10 8 6 4 2 0

0

10

0

Relative fold change

Ly6d

Relative fold change

Relative fold change

Afp

5

*

0

15

10

Spink1

Relative fold change

120

Relative fold change

Relative fold change

Tff3

0

B

150

250 140

Relative fold change

A

6

Plaur

*

Gpc3

*

(A) Expression of genes that are frequently induced in mouse and human HCC in Akt1hep / ;Akt2 / livers versus Akt2 / livers of 6-month old male mice was measured by qRT-PCR (n = 4, ±SEM, *p % 0.05, **p < 0.005). (B) The expression of Fosl2 and associated target genes in the liver of 6-month old male Akt1hep / ; Akt2 / and Akt2 / mice was determined by qRTPCR (n = 4, ±SEM, *p < 0.05). (C) Expression of Igf2bp1 mRNA in Akt1hep / ; Akt2 / liver compared with Akt2 / liver (n = 3, ±SEM, *p = 0.052). (D) Expression of Igf2bp3 mRNA in Akt1hep / ; Akt2 / liver compared with Akt2 / liver (n = 4, ±SEM, **p < 0.005). (E) Expression of IGF2BP3 protein in Akt1hep / ; Akt2 / livers compared with Akt2 / livers. See also Figure S8.

4

Relative fold change

Relative fold change

determine the role of Akt1 and Akt2 as well as insulin levels in HCC, 2-week-old 0 0 0 Akt1 / and Akt2 / male mice were Akt2-/subjected to intraperitoneal injection of Akt1hep-/-;Akt2-/C D E 5 ** 3000 Igf2bp1 * 25 mg/ml DEN and hepatic tumor inciIgf2bp3 2500 4 2000 dence was analyzed 12 months later. 20 IGF2BP3 3 Unexpectedly, the incidences of macro15 scopic tumors in Akt1 / and Akt2 / 2 β-actin 10 mice are not significantly different from 1 5 that of WT mice (Figure 7A). Notably, Akt2 0 0 is the major Akt isoform expressed in the liver, yet its deletion resulted in not only cancers, including HCC, in which they have been associated no resistance to hepatocarcinogenesis but also a rather marked with the induction of cell proliferation, cell survival, and invasive- increase in the incidence of metastasis. A high incidence of lung ness (Bell et al., 2013; Lederer et al., 2014). Both IGF2BP1 and tumors was observed in Akt2 / mice, wherein 50% of the mice IGF2BP3 (also known as IMP3) are highly and frequently ex- showed macroscopic lung tumors, with a total of 12 lung tumors pressed in human HCC, and their expression is correlated with in 16 mice (Figures 7B and 7C). By comparison, only 10% of the aggressiveness, invasiveness, and poor prognosis (Gutschner WT mice developed macroscopic lung tumors, with a total of et al., 2014; Jeng et al., 2008; Wachter et al., 2012). Because three tumors in 20 mice. This difference was much more of its selective expression in aggressive HCC, IGF2BP3 was profound when microscopic lesions in the lung were quantified suggested as a prognostic marker (Wachter et al., 2012). Thus, (Figure 7D; WT mice). These tumors were derived from primary based on studies in human HCC, the tumors developed in the liver tumors based on pathology (Figure 7E), and they expressed a-fetoprotein (Afp), which was not expressed in the lung but only livers of Akt1hep / ;Akt2 / mice are considered aggressive. Together, the results show that the HCC is initiated by cell in liver tumors (Figure 7F). Therefore, we concluded that these death of hepatocytes in which both Akt1 and Akt2 are deleted, tumors were metastases that had originated from primary liver resulting in liver injury followed by inflammation, which are tumors. The major physiological difference between Akt2 / mice and FoxO1-dependent. The liver injury, inflammation, and high IL-6 level are considered to be prerequisites for HCC development. either Akt1 / or WT mice is the insulin resistance and high blood IL-6 activates STAT3 and promotes compensatory proliferation insulin level in Akt2 / mice. We found up to 10-fold higher level of in hepatocytes that have escaped cell death, subsequently pro- insulin in Akt2 / mice (Chen et al., 2009). Indeed, the Akt2 / moting tumor development. In addition, based on gene expres- mice that developed metastases to the lung had extremely high sion, Akt1hep / ;Akt2 / liver tumors appeared to be character- levels of serum insulin compared with the WT mice that did not ized as undifferentiated and of embryonic origin, and the high develop lung metastasis (Figure 7G). The high level of insulin in expression levels of genes, such as IGF2BP1 and, particularly, Akt2 / mice leads to a compensatory activation of Akt1 (Xu IGF2BP3, facilitated tumorigenesis. Finally, autocrine secretion et al., 2012) (Figure 7H), as well as Erk1/2 activation (Figure 7I), that could be pro-metastatic. Since Akt2 is the major Akt isoform of IL-6 could further facilitate HCC development. It is possible that the high serum insulin level in Akt1hep / ; expressed in the liver and its deletion reduced about two thirds of Akt2 / mice also facilitated tumor development. To further total hepatic Akt activity, these results further support the notion 2

8 Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc.

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

that inhibiting hepatic Akt activity accelerates HCC progression. Thus, the inhibition of hepatic Akt activity induces liver injury and inflammation in a FoxO1-dependent manner, which in turn induces compensatory proliferation via STAT3 activation. Subsequently, additional acquired mutations lead to HCC (Figure 8A). A high level of insulin may facilitate the development of HCC. Thus, the deficiency of Akt2 only, which induces high levels of insulin, facilitates HCC progression and metastasis (Figure 8B). Consequently, the results could explain, at least in part, why fatty liver that inhibits hepatic Akt activity facilitates HCC. DISCUSSION Because Akt is frequently activated in human cancers and inhibitors of Akt are being used in clinical trials for cancer therapy, it has become necessary to understand the consequences of the deletion of Akt isoforms on the survival and physiology of adult mice. To address this important issue, we systemically deleted Akt isoforms in adult mice. In the present study, we showed that deletion of both Akt1 and Akt3 in adult mice was tolerated, while deletion of both Akt1 and Akt2 elicited rapid mortality. The rapid mortality was preceded by a transient increase followed by a decrease in glucose and insulin levels and subsequent cachexia and hypoglycemia. Our results indicated that the mice developed tissue damage and inflammation. The tissue damage disrupted the villi in the intestine, which impaired food absorption. The mice lost body fat possibly because of its utilization as an alternative source of nutrients. The most intriguing and unexpected observation that was uncovered in our studies is the early onset of spontaneous hepatocarcinogenesis in Akt1hep / Akt2 / mice with complete penetrance. Akt is perhaps the most frequently activated oncoprotein in human cancers, and activation of Akt was implicated in the genesis of HCC (Moeini et al., 2012; Shearn and Petersen, 2015). Furthermore, hepatic deletion of Pten in mice induces HCC (Horie et al., 2004). Inhibitors of Akt and PI3K are being used for cancer therapy (Dienstmann et al., 2014; Fruman and Rommel, 2014), and the germline deletion of Akt1 in mice was reported to significantly impair tumor development in several mouse models of cancer (Chen et al., 2006; Hollander et al., 2011; Ju et al., 2007; Maroulakou et al., 2007; Skeen et al., 2006). Moreover, systemic deletion of Akt1 after tumor onset in Trp53 / mice halts and regresses thymic lymphoma (Yu et al., 2015). By contrast, our results showed that hepatic inactivation of Akt1 and Akt2 induces HCC with markedly faster onset than by the activation of Akt through hepatic Pten deletion, which occurs 75 weeks after birth (Horie et al., 2004). Although this observation may constitute an apparent paradox, it is consistent with the role of obesity, and a high fat diet (HFD) in HCC. It has been documented that obesity and a fatty liver can promote human HCC, and in mouse models, HFD has been shown to accelerate HCC (Baffy et al., 2012; Park et al., 2010; Sun and Karin, 2012). Under these conditions, HCC develops despite insulin resistance and the inhibition of hepatic Akt activity. The contribution of HFD and fatty liver to HCC genesis has been attributed to oxidative and ER stress (Nakagawa et al., 2014). However, our results suggest that hepatic Akt inhibition alone, as a consequence of obesity, can induce liver damage and inflammation

and subsequently HCC, adding a potential mechanism of protumorigenesis in the fatty liver. Notably, the hepatic deletion of Tsc1 in mice, which inhibits hepatic Akt activity, induces spontaneous HCC albeit with slower kinetics compared with what we observed in Akt1hep / Akt2 / mice (Menon et al., 2012). The development of HCC in these mice was attributed to the hyperactivation of mTORC1 (Menon et al., 2012). However, based on our studies, it is possible that hepatic Akt inactivation in this mouse model also contributed to HCC. HCC development is often preceded by liver injury and inflammation and, consequently, the production of pro-tumorigenic cytokines, in particular IL-6, which induces STAT3 activation in hepatocytes. Consistently, we observed hepatic cell death and liver injury, as well as inflammation and elevated circulating IL-6 levels, in liver-tumor-bearing Akt1hep / ;Akt2 / mice. The liver injury and inflammation appeared to be largely FoxO1-dependent because the hepatic deletion of both Akt1 and Akt2 induced liver injury and inflammation that were blunted when FoxO1 was also deleted. Indeed, although Akt2 is the major isoform expressed in the liver (Chen et al., 2006), deletion / of Akt1 in the liver of Akt2 mice further increased FoxO target gene expression. This phenomenon could be explained by the high level of insulin in Akt2 / mice, which hyperactivates hepatic Akt1 in these mice and therefore causes a reduction in total hepatic Akt activity that is lower than that anticipated for the full activation of hepatic FoxO. Following the deletion of hepatic Akt1 in Akt2 / mice, the reduction in total Akt activity becomes sufficient to fully activate FoxO. Thus, the high activity of FoxO could induce apoptosis, at least in part, by elevating the expression of the pro-apoptotic genes, Fasl and Bcl2l11. Interestingly, apoptosis largely occurred in hepatocytes located outside or distal to the area of the tumor. The dying hepatocytes elicit inflammation by macrophages and other cells, which in turn results in increased IL-6 and other pro-tumorigenic cytokines. IL-6 then activates STAT3 in the surviving hepatocytes that subsequently give rise to the tumors. Although our results support this mechanism, we cannot completely exclude other not mutually exclusive mechanisms that contribute to the rapid onset of HCC in Akt1hep / ;Akt2 / mice. It is worth noting, for example, that mice with hepatic deletion of Stat5 are more susceptible to hepatocarcinogenesis, and that the hepatic deletion of Stat5 induces STAT3 activity (Cui et al., 2007). Thus, it is possible that STAT5 inactivation in the livers of Akt1hep / ;Akt2 / mice contributed to the increased STAT3 activity and hepatocarcinogenesis. Notably, we found that the source of IL-6 in Akt1hep / ; Akt2 / livers was not necessarily only macrophages but also certain hepatocytes that begin to express IL-6. Interestingly, the IL-6-positive cells were usually located at a site that was distant from the tumor site. Although the origin of these cells is not clear, it is possible that they evolved as a consequence of the Akt1 and Akt2 deletion to maintain survival via IL-6dependent and Akt-independent mechanisms. Autocrine secretion of IL-6 has also been observed in human HCC, and a recent study showed that in DEN-induced HCC in mice, certain progenitor cells start expressing IL-6 to promote their malignant progression (He et al., 2013). The histopathology revealed two types of tumors that were morphologically consistent with clear cell HCC and Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc. 9

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

A

B

C D E

F

G

H

I

Figure 7. DEN-Induced Hepatocarcinogenesis and Lung Metastasis in Akt1–/– and Akt2–/– Mice (A) The incidence of macroscopic tumors R1 cm in diameter in the livers of Akt1 / , Akt2 / , and WT mice, 12 months after exposure to DEN. (B) The incidence of macroscopic metastatic tumors in the lungs of Akt1 / , Akt2 / , and WT mice 12 months after exposure to DEN (results are presented as the mean ± SEM). (C) Images depicting lungs with macroscopic lung tumors indicated by arrows. Scale bar, 10 mm. (D) The number of microscopic metastatic tumors in the lung of Akt2 / , and WT mice, 12 months after exposure to DEN. Data are presented as the mean ± SEM. (legend continued on next page)

10 Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc.

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

A Compensatory Proliferation + Survival

Insulin FOXO

HCC

Inflammation

IL-6

STAT3 IL-6, TNF

Cell Death

Other mutations Igf2bp3 Fos Fosl2 Cd44

5-6 Months

B Insulin Cell Death

Compensatory Inflammation Proliferation + Survival

HCC

STAT3

DEN

trabecular HCC in Akt1hep / ;Akt2 / livers. Gene expression analysis demonstrated the expression of many genes in the Akt1hep / ;Akt2 / liver that are also induced in human HCC and in a mouse model of DEN-induced HCC (He et al., 2013; Lee et al., 2006). The gene expression analysis showed that several embryonic genes that are not normally expressed in the adult liver are induced in the tumor-bearing Akt1hep / ;Akt2 / liver. Of particular interest is the expression of the network of genes associated with the induction of Fos, Fosl1, and Fosl2, which have been implicated in driving tumorigenesis in an aggressive subtype of human HCC (Lee et al., 2006). Among the embryonic genes that were strongly expressed in the Akt1hep / ;Akt2 / liver were Igf2bp1 and Igf2bp3. Both IGF2BP1 and IGF2BP3 are highly and frequently expressed in human HCC, and their expression has been correlated to aggressiveness, invasiveness, and a poor prognosis (Gutschner et al., 2014; Jeng et al., 2008; Wachter et al., 2012). Because of its selective expression in aggressive HCC, IGF2BP3 has been suggested to be a prognostic marker (Wachter et al., 2012). Thus, based on studies in human HCC, the tumors that develop in the livers of Akt1hep / ;Akt2 / mice are considered aggressive. Finally, we found that neither Akt1 / nor Akt2 / mice were resistant to DEN-induced HCC. Importantly, Akt2 / mice displayed a marked increase in the incidence of lung metastases. We attributed the high incidence of metastasis to the very high level of circulating insulin in Akt2 / mice. As shown previ-

Metastasis

Figure 8. Illustration Depicting HCC Development and Progression in Akt1hep–/–; Akt2–/– and Akt2–/– Mice (A) The deletion of hepatic Akt1 and Akt2 induces liver damage and apoptosis, which is largely dependent on the activation of FoxO. FoxO induces inflammation and IL-6 and TNF-a production by macrophages and plasma cells. IL-6 induces compensatory proliferation by activating STAT3 in the survived hepatocytes. Eventually, these cells give rise to HCC with gene signatures similar to human HCC. Autocrine production of IL-6 by hepatocytes was also observed. The high level of insulin in the mice may facilitate HCC progression. (B) Exposure of Akt2 / mice that display a high level of insulin to DEN-induced hepatocarcinogenesis induces a high incidence of metastasis to the lung.

12 Months

ously, the high level of insulin in Akt2 / mice could hyperactivate other Akt isoforms and potentially other downstream signaling pathways in different tissues (Xu et al., 2012). The activation of these signaling pathways could contribute to the high incidence of metastasis in Akt2 / mice. However, we cannot completely exclude the possibility that the deletion of Akt2 in other tissues facilitate metastasis in a non-cell autonomous manner. Taken together, the results presented here provided an unexpected conceptual paradigm whereby Akt activity, which is required for cancer development in general, is not required for the initiation of liver cancer, and its inhibition could promote liver cancer and liver cancer progression. Our results suggest that systemic and efficient inhibition of both Akt1 and Akt2 by panAkt inhibitors might not be well tolerated. Furthermore, because orally or systemically delivered drugs can accumulate at the highest levels in the liver, they could inhibit the hepatic activity of Akt to a similar extent to that observed for the hepatic deletion of Akt1 and Akt2. Unlike genetic deletion, drug therapy is not usually administered for an unlimited time period. However, our results suggest that the ablation of Akt1 and Akt2 could elicit a predisposition to HCC by inducing liver damage and inflammation. Because liver damage and inflammation could occur even after transient inhibition of hepatic Akt activity, it would subsequently increase the risk for liver cancer. This process could be exacerbated by obesity, which also induces insulin resistance and inhibits hepatic Akt activity. When we treated the mice

(E) H&E stained lung sections from WT, Akt1 / , and Akt2 / mice. Tumor is marked in the Akt2 / lung section (Left panel-scale bar 500mM; Right panel-scale bar 100mM). (F) RT-PCR analysis of Afp expression in tumors developed in the lung of Akt2 / mice. (G) Representative images of livers bearing tumors and corresponding lungs (tumors are indicated by arrowheads) from DEN treated Akt2 / and WT mice (scale bars, 10 mm). Total number of macroscopic liver tumors and number of tumors R1 cm in diameter are shown in parentheses. Fed glucose and insulin levels in the corresponding mice are shown adjacent to the images. (H) Immunoblot showing p-Akt1 in tumors derived from either WT or Akt2 / livers. (I) Immunoblot showing p-ERK in tumors derived from either WT or Akt2 / livers. Upper panel, total Akt (t-Akt); middle panel, p-ERK; bottom panel, total ERK (t-ERK).

Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc. 11

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

with a moderately high dose of the pan-Akt inhibitor MK2206, we recapitulated the consequences of systemic Akt1 and Akt2 deletion. However, the mice recovered after the treatment was stopped. We followed these mice for 6 months, and thus far we did not find liver tumors, although we cannot completely exclude the possibility that liver tumors will eventually develop. The high systemic insulin levels following pan-Akt inhibition or Akt2 inhibition could counteract the benefits of cancer therapy, due to the hyperactivation of other pro-tumorigenic signaling pathways downstream of insulin. Because HCC development is characterized by a long latency period, HCC after treatment with pan-Akt inhibitors might not be manifested early but rather late after treatment. EXPERIMENTAL PROCEDURES Mice Akt1 / , Akt2 / , and Akt3 / mice have been previously described (Chen et al., 2001, 2009; Peng et al., 2003). Akt1f/f mice were generated by Ozgene (Australia), and the generation of Akt1f/f;R26RCreERT2 mice has been described elsewhere (Yu et al., 2015). Akt1f/f;Akt2f/f and Akt1f/f;Akt2f/f; FoXo1f/f mice have been previously described (Lu et al., 2012). Akt1f/f;Akt2 / ; R26CreERT2 and Akt1f/f;Akt3 / ;R26CreERT2 mice were generated by crossing Akt1f/f;R26CreERT2 with either Akt2 / or Akt3 / mice. R26CreERT2 knockin mice (strain 01XAB) were obtained from Tyler Jacks’ laboratory (Massachusetts Institute of Technology). Akt1hep / ;Akt2 / mice were generated by crossing Akt1/f/f;Akt2 / with the albumin-CRE mouse strain (obtained from the Jackson Laboratory). All mice are in the C57BL6 background. The generation of Akt1f/f;Akt2f/f;R26CreERT2 was performed by crossing Akt1f/f; R26CreERT2 in the FVB background with Akt2f/f mice in the FVB background. Systemic injection of AAV into Akt1f/f;Akt2f/f or Akt1f/f;Akt2f/f;Foxo1f/f mice was conducted as previously described (Lu et al., 2012). Tamoxifen injection for systemic activation of Cre recombinase has been previously described (Yu et al., 2015). All animal experiments were approved by the University of Illinois at Chicago Institutional Animal Care and Use Committee, as required by United States Animal Welfare Act, and the NIH’s policy. Histopathology, Immunocytochemistry, and Immunofluorescence Tissue samples were fixed in 10% formalin overnight and then processed and embedded in paraffin. The sections (5 mm) were stained with H&E. For immunohistochemistry, the tissue sections were incubated at 95 C in 0.01 M citric acid (pH 6.0) for 20 min, followed by a 30-min cool down. The sections were then treated with 0.3% hydrogen peroxide to block endogenous peroxides and then washed with PBS and blocked with serum. The samples were incubated with primary antibodies overnight at 4 C, followed by incubation with the biotinylated secondary antibody and avidin-biotin complexes. Signals were visualized with diaminobenzidine. Some of the sections were lightly counterstained with hematoxylin. The primary antibodies included anti-Ki67 (Cell Signaling), anti-F4/80 (Abcam), anti-cleaved caspase 3 (Cell Signaling), and anti-IL-6 (Abcam). Biotin-conjugated secondary antibody kits, avidin-biotin complexes, and diaminobenzidine were purchased from Vector Laboratories. Ki67, cleaved caspase-3, and F4/80-positive cells were counted in ten fields at 2003 magnification per slide using ImageJ software (NIH). For immunofluorescence, slides were blocked with 10% goat serum in PBS for 30 min and incubated with rabbit anti-AKT1 (Abcam ab32505, lot GR145853-2) and Rat anti-F4/80 (Abcam ab6640, lot GR169072-1) at 4 C overnight. After washing in TNT buffer (0.1 M Tris-HCl [pH 7.5], 150 mM NaCl, 0.05% Tween 20), slides were incubated with tetramethylrhodamineconjugated anti-rabbit for Akt1 and fluorescein isothiocyanate anti-rat for F4/80 secondary antibodies, and then stained with 1 mg/ml DAPI for 10 min at room temperature. Slides were mounted with mounting medium from Vector Laboratories. Slides were examined and imaged by a Zeiss LSM 700 confocal microscope using the manufacturer’s imaging software ZEN (ZEISS Efficient Navigation). Other detailed experimental procedures are described in the Supplemental Experimental Procedures.

12 Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc.

ACCESSION NUMBERS The accession number for the RNA sequencing data reported in this manuscript is GEO: GSE77862. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures and eight figures and can be found with this article online at http://dx.doi. org/10.1016/j.ccell.2016.02.008. AUTHOR CONTRIBUTIONS Q.W., W.N.Y., X.D.P., X.C., and N.H. designed the experiments and analyzed the data. W.N.Y. generated mice and conducted the initial experiments. Q.W. and X.C. completed the experiments for publication. X.C. generated the mice with systemic Akt1 and Akt2 deletion and conducted experiments with these mice. X.D.P. conducted the hepatocarcinogenesis experiments. S.M.J. generated the Akt3 / mice with systemic deletion of Akt1 and analyzed them. G.G. made the pathological evaluations of the tissue sections. M.J.B provided some of the mice for the experiments. Q.W., W.N.Y., and N.H. wrote the paper. ACKNOWLEDGMENTS These studies were supported by NIH grants R01AG016927 and R01CA090764, a VA Merit Award BX000733 to N.H., and an NIH grant RO1 DK56886 to M.J.B. M.J.B. is an employee of Pfizer Inc. Received: August 4, 2015 Revised: December 12, 2015 Accepted: February 16, 2016 Published: March 17, 2016 REFERENCES Baffy, G., Brunt, E.M., and Caldwell, S.H. (2012). Hepatocellular carcinoma in non-alcoholic fatty liver disease: an emerging menace. J. Hepatol. 56, 1384– 1391. Bell, J.L., Wachter, K., Muhleck, B., Pazaitis, N., Kohn, M., Lederer, M., and Huttelmaier, S. (2013). Insulin-like growth factor 2 mRNA-binding proteins (IGF2BPs): post-transcriptional drivers of cancer progression? Cell Mol. Life Sci. 70, 2657–2675. Chen, W.S., Xu, P.Z., Gottlob, K., Chen, M.L., Sokol, K., Shiyanova, T., Roninson, I., Weng, W., Suzuki, R., Tobe, K., et al. (2001). Growth retardation and increased apoptosis in mice with homozygous disruption of the Akt1 gene. Genes Dev. 15, 2203–2208. Chen, M.L., Xu, P.Z., Peng, X., Chen, W.S., Guzman, G., Yang, X., Di Cristofano, A., Pandolfi, P.P., and Hay, N. (2006). The deficiency of Akt1 is sufficient to suppress tumor development in Pten+/- mice. Genes Dev. 20, 1569– 1574. Chen, W.S., Peng, X.D., Wang, Y., Xu, P.Z., Chen, M.L., Luo, Y., Jeon, S.M., Coleman, K., Haschek, W.M., Bass, J., et al. (2009). Leptin deficiency and beta-cell dysfunction underlie type 2 diabetes in compound Akt knockout mice. Mol. Cell Biol. 29, 3151–3162. Cui, Y., Hosui, A., Sun, R., Shen, K., Gavrilova, O., Chen, W., Cam, M.C., Gao, B., Robinson, G.W., and Hennighausen, L. (2007). Loss of signal transducer and activator of transcription 5 leads to hepatosteatosis and impaired liver regeneration. Hepatology 46, 504–513. Dienstmann, R., Rodon, J., Serra, V., and Tabernero, J. (2014). Picking the point of inhibition: a comparative review of PI3K/AKT/mTOR pathway inhibitors. Mol. Cancer Ther. 13, 1021–1031. Dong, X.C., Copps, K.D., Guo, S., Li, Y., Kollipara, R., DePinho, R.A., and White, M.F. (2008). Inactivation of hepatic Foxo1 by insulin signaling is required for adaptive nutrient homeostasis and endocrine growth regulation. Cell Metab. 8, 65–76.

Please cite this article in press as: Wang et al., Spontaneous Hepatocellular Carcinoma after the Combined Deletion of Akt Isoforms, Cancer Cell (2016), http://dx.doi.org/10.1016/j.ccell.2016.02.008

Fruman, D.A., and Rommel, C. (2014). PI3K and cancer: lessons, challenges and opportunities. Nat. Rev. Drug Discov. 13, 140–156. Gutschner, T., Hammerle, M., Pazaitis, N., Bley, N., Fiskin, E., Uckelmann, H., Heim, A., Grobeta, M., Hofmann, N., Geffers, R., et al. (2014). Insulin-like growth factor 2 mRNA-binding protein 1 (IGF2BP1) is an important protumorigenic factor in hepatocellular carcinoma. Hepatology 59, 1900–1911. Hay, N. (2011). Akt isoforms and glucose homeostasis - the leptin connection. Trends Endocrinol. Metab. 22, 66–73. He, G., Dhar, D., Nakagawa, H., Font-Burgada, J., Ogata, H., Jiang, Y., Shalapour, 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. Heindryckx, F., Colle, I., and Van Vlierberghe, H. (2009). Experimental mouse models for hepatocellular carcinoma research. Int. J. Exp. Pathol. 90, 367–386. Herrington, J., Smit, L.S., Schwartz, J., and Carter-Su, C. (2000). The role of STAT proteins in growth hormone signaling. Oncogene 19, 2585–2597. Hollander, M.C., Maier, C.R., Hobbs, E.A., Ashmore, A.R., Linnoila, R.I., and Dennis, P.A. (2011). Akt1 deletion prevents lung tumorigenesis by mutant K-ras. Oncogene 30, 1812–1821. Horie, Y., Suzuki, A., Kataoka, E., Sasaki, T., Hamada, K., Sasaki, J., Mizuno, K., Hasegawa, G., Kishimoto, H., Iizuka, M., et al. (2004). Hepatocyte-specific Pten deficiency results in steatohepatitis and hepatocellular carcinomas. J. Clin. Invest. 113, 1774–1783. Jeng, Y.M., Chang, C.C., Hu, F.C., Chou, H.Y., Kao, H.L., Wang, T.H., and Hsu, H.C. (2008). RNA-binding protein insulin-like growth factor II mRNA-binding protein 3 expression promotes tumor invasion and predicts early recurrence and poor prognosis in hepatocellular carcinoma. Hepatology 48, 1118–1127. Ju, X., Katiyar, S., Wang, C., Liu, M., Jiao, X., Li, S., Zhou, J., Turner, J., Lisanti, M.P., Russell, R.G., et al. (2007). Akt1 governs breast cancer progression in vivo. Proc. Natl. Acad. Sci. USA 104, 7438–7443. Klammt, J., Pfaffle, R., Werner, H., and Kiess, W. (2008). IGF signaling defects as causes of growth failure and IUGR. Trends Endocrinol. Metab. 19, 197–205. Lederer, M., Bley, N., Schleifer, C., and Huttelmaier, S. (2014). The role of the oncofetal IGF2 mRNA-binding protein 3 (IGF2BP3) in cancer. Semin. Cancer Biol. 29, 3–12. Lee, J.S., Heo, J., Libbrecht, L., Chu, I.S., Kaposi-Novak, P., Calvisi, D.F., Mikaelyan, A., Roberts, L.R., Demetris, A.J., Sun, Z., et al. (2006). A novel prognostic subtype of human hepatocellular carcinoma derived from hepatic progenitor cells. Nat. Med. 12, 410–416. Leung, K.C., Doyle, N., Ballesteros, M., Waters, M.J., and Ho, K.K. (2000). Insulin regulation of human hepatic growth hormone receptors: divergent effects on biosynthesis and surface translocation. J. Clin. Endocrinol. Metab. 85, 4712–4720. Li, Y., Tang, Z.Y., and Hou, J.X. (2012). Hepatocellular carcinoma: insight from animal models. Nat. Rev. Gastroenterol. Hepatol. 9, 32–43. Lu, M., Wan, M., Leavens, K.F., Chu, Q., Monks, B.R., Fernandez, S., Ahima, R.S., Ueki, K., Kahn, C.R., and Birnbaum, M.J. (2012). Insulin regulates liver metabolism in vivo in the absence of hepatic Akt and Foxo1. Nat. Med. 18, 388–395. Luedde, T., Beraza, N., Kotsikoris, V., van Loo, G., Nenci, A., De Vos, R., Roskams, T., Trautwein, C., and Pasparakis, M. (2007). Deletion of NEMO/ IKKgamma in liver parenchymal cells causes steatohepatitis and hepatocellular carcinoma. Cancer Cell 11, 119–132. Maroulakou, I.G., Oemler, W., Naber, S.P., and Tsichlis, P.N. (2007). Akt1 ablation inhibits, whereas Akt2 ablation accelerates, the development of mammary adenocarcinomas in mouse mammary tumor virus (MMTV)-ErbB2/neu and MMTV-polyoma middle T transgenic mice. Cancer Res. 67, 167–177.

Menon, S., Yecies, J.L., Zhang, H.H., Howell, J.J., Nicholatos, J., Harputlugil, E., Bronson, R.T., Kwiatkowski, D.J., and Manning, B.D. (2012). Chronic activation of mTOR complex 1 is sufficient to cause hepatocellular carcinoma in mice. Sci. Signal. 5, ra24. Moeini, A., Cornella, H., and Villanueva, A. (2012). Emerging signaling pathways in hepatocellular carcinoma. Liver Cancer 1, 83–93. 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. Nogueira, V., Sundararajan, D., Kwan, J.M., Peng, X.D., Sarvepalli, N., Sonenberg, N., and Hay, N. (2012). Akt-dependent Skp2 mRNA translation is required for exiting contact inhibition, oncogenesis, and adipogenesis. EMBO J. 31, 1134–1146. 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. Peng, X.D., Xu, P.Z., Chen, M.L., Hahn-Windgassen, A., Skeen, J., Jacobs, J., Sundararajan, D., Chen, W.S., Crawford, S.E., Coleman, K.G., and Hay, N. (2003). Dwarfism, impaired skin development, skeletal muscle atrophy, delayed bone development, and impeded adipogenesis in mice lacking Akt1 and Akt2. Genes Dev. 17, 1352–1365. Shearn, C.T., and Petersen, D.R. (2015). Understanding the tumor suppressor PTEN in chronic alcoholism and hepatocellular carcinoma. Adv. Exp. Med. Biol. 815, 173–184. Shin, D.J., Joshi, P., Hong, S.H., Mosure, K., Shin, D.G., and Osborne, T.F. (2012). Genome-wide analysis of FoxO1 binding in hepatic chromatin: potential involvement of FoxO1 in linking retinoid signaling to hepatic gluconeogenesis. Nucleic Acids Res. 40, 11499–11509. Skeen, J.E., Bhaskar, P.T., Chen, C.C., Chen, W.S., Peng, X.D., Nogueira, V., Hahn-Windgassen, A., Kiyokawa, H., and Hay, N. (2006). Akt deficiency impairs normal cell proliferation and suppresses oncogenesis in a p53-independent and mTORC1-dependent manner. Cancer Cell 10, 269–280. Sun, B., and Karin, M. (2012). Obesity, inflammation, and liver cancer. J. Hepatol. 56, 704–713. Taniguchi, K., and Karin, M. (2014). IL-6 and related cytokines as the critical lynchpins between inflammation and cancer. Semin. Immunol. 26, 54–74. Umemura, A., Park, E.J., Taniguchi, K., Lee, J.H., Shalapour, S., Valasek, M.A., Aghajan, M., Nakagawa, H., Seki, E., Hall, M.N., and Karin, M. (2014). Liver damage, inflammation, and enhanced tumorigenesis after persistent mTORC1 inhibition. Cell Metab. 20, 133–144. Ventura, A., Kirsch, D.G., McLaughlin, M.E., Tuveson, D.A., Grimm, J., Lintault, L., Newman, J., Reczek, E.E., Weissleder, R., and Jacks, T. (2007). Restoration of p53 function leads to tumour regression in vivo. Nature 445, 661–665. Wachter, D.L., Kristiansen, G., Soll, C., Hellerbrand, C., Breuhahn, K., Fritzsche, F., Agaimy, A., Hartmann, A., and Riener, M.O. (2012). Insulin-like growth factor II mRNA-binding protein 3 (IMP3) expression in hepatocellular carcinoma. A clinicopathological analysis with emphasis on diagnostic value. Histopathology 60, 278–286. Xu, P.Z., Chen, M.L., Jeon, S.M., Peng, X.D., and Hay, N. (2012). The effect Akt2 deletion on tumor development in Pten(+/-) mice. Oncogene 31, 518–526. Yu, W.N., Nogueira, V., Sobhakumari, A., Patra, K.C., Bhaskar, P.T., and Hay, N. (2015). Systemic Akt1 deletion after tumor onset in p53(-/-) mice increases lifespan and regresses thymic lymphoma emulating p53 restoration. Cell Rep. 12, 610–621.

Cancer Cell 29, 1–13, April 11, 2016 ª2016 Elsevier Inc. 13