Astrocyte elevated gene-1 and c-Myc cooperate to promote hepatocarcinogenesis in mice


  • Potential conflict of interest: Nothing to report.

  • The present study was supported, in part, by grants from the James S. McDonnell Foundation and National Cancer Institute (NCI; grant nos.: R01 CA138540 [to D.S.] and R01 CA134721 [to P.B.F.]). The VCU Massey Cancer Center Transgenic/Knockout Mouse Facility, supported, in part, with funding from the National Institutes of Health/NCI Cancer Center (Support Grant P30 CA 016059), provided services in support of this project. P.B.F. holds the Thelma Newmeyer Corman Chair in Cancer Research. D.S. is the Harrison Endowed Scholar in Cancer Research and a Blick scholar.

  • See Editorial on Page 757


Astrocyte elevated gene-1 (AEG-1) and c-Myc are overexpressed in human hepatocellular carcinoma (HCC) functioning as oncogenes. AEG-1 is transcriptionally regulated by c-Myc, and AEG-1 itself induces c-Myc by activating the Wnt/β-catenin–signaling pathway. We now document the cooperation of AEG-1 and c-Myc in promoting hepatocarcinogenesis by analyzing hepatocyte-specific transgenic mice expressing either AEG-1 (albumin [Alb]/AEG-1), c-Myc (Alb/c-Myc), or both (Alb/AEG-1/c-Myc). Wild-type and Alb/AEG-1 mice did not develop spontaneous HCC. Alb/c-Myc mice developed spontaneous HCC without distant metastasis, whereas Alb/AEG-1/c-Myc mice developed highly aggressive HCC with frank metastasis to the lungs. Induction of carcinogenesis by N-nitrosodiethylamine significantly accelerated the kinetics of tumor formation in all groups. However, in Alb/AEG-1/c-Myc, the effect was markedly pronounced with lung metastasis. In vitro analysis showed that Alb/AEG-1/c-Myc hepatocytes acquired increased proliferation and transformative potential with sustained activation of prosurvival and epithelial-mesenchymal transition–signaling pathways. RNA-sequencing analysis identified a unique gene signature in livers of Alb/AEG-1/c-Myc mice that was not observed when either AEG-1 or c-Myc was overexpressed. Specifically, Alb/AEG-1/c-Myc mice overexpressed maternally imprinted noncoding RNAs (ncRNAs), such as Rian, Meg-3, and Mirg, which are implicated in hepatocarcinogenesis. Knocking down these ncRNAs significantly inhibited proliferation and invasion by Alb/AEG-1/c-Myc hepatocytes. Conclusion: Our studies reveal a novel cooperative oncogenic effect of AEG-1 and c-Myc that might explain the mechanism of aggressive HCC. Alb/AEG-1/c-Myc mice provide a useful model to understand the molecular mechanism of cooperation between these two oncogenes and other molecules involved in hepatocarcinogenesis. This model might also be of use for evaluating novel therapeutic strategies targeting HCC. (Hepatology 2015;61:915–929)


astrocyte elevated gene 1




protein kinase B




B-cell lymphoma 2


B-cell lymphoma-extra large




complementary DNA


colony formation assay


cellular inhibitor of apoptosis protein 1




deiodinase, iodothyronine, type III


delta-like 1




extracellular matrix


epidermal growth factor


epidermal growth factor receptor


epithelial-mesenchymal transition


extracellular signal-regulated kinase


hepatocellular carcinoma


mitogen-activated protein kinase




3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide


noncoding RNAs


nuclear factor kappa B


National Institutes of Health


proliferating cell nuclear antigen


promyelocytic leukemia zinc finger


RNA sequencing


reads per kilobase per million


reverse-transcriptase polymerase chain reaction


standard error of the mean


small interfering RNAs


mothers against decapentaplegic


secreted protein acidic and rich in cysteine


signal transducer and activator of transcription 3


transcription factor 8




transforming growth factor beta


wild type


X-linked inhibitor of apoptosis protein


zona occludens 1

Astrocyte elevated gene-1 (AEG-1), also known as metadherin or LYRIC, is overexpressed in all cancers studied to date.[1] In hepatocellular carcinoma (HCC), a high percentage of patients demonstrate AEG-1 overexpression in both messenger RNA and protein levels.[2, 3] In HCC patients, AEG-1 expression is closely associated with microvascular invasion, pathological satellites, tumor differentiation, and tumor node metastatis stage and both overall survival and cumulative recurrence rates show an inverse correlation with AEG-1 expression levels.[3, 4] Gain- and loss-of-function studies using human HCC cell lines confirmed that AEG-1 confers highly aggressive, angiogenic, metastatic, and chemoresistance phenotypes.[2, 5-7] A transgenic (Tg) mouse with hepatocyte-specific overexpression of AEG-1 (albumin [Alb]/AEG-1) did not develop spontaneous HCC; however, upon treatment with the chemical carcinogen, N-nitrosodiethylamine (DEN), HCC developed with a significantly accelerated kinetics with increased angiogenesis and chemoresistance.[8] Collectively, these studies indicate that although AEG-1 itself may not initiate HCC, it plays a key role in HCC progression and metastasis.

Multiple mechanisms underlie AEG-1 overexpression in cancer, which include amplification of the AEG-1 locus, 8q22, regulation by multiple tumor-suppressor microRNAs (miRNAs), and post-translational regulation by monoubiquitination, that increase stability of AEG-1 protein.[2, 9-14] The oncogenic transcription factor, c-Myc, directly binds to the AEG-1 promoter and regulates its transcription.[15] Overexpression of c-Myc is detected in a high percentage of HCC patients[16, 17] and thus might be a key mechanism by which AEG-1 expression is induced in HCC. Gain of chromosome 8q is a defining feature of human HCC leading to coamplification of AEG-1 and c-Myc, the latter located at 8q24.1.[18] On the other hand, AEG-1 itself induces c-Myc expression by activating the Wnt/β-catenin–signaling pathway.[2] AEG-1 also interacts with promyelocytic leukemia zinc finger (PLZF), a transcriptional repressor, inhibiting its ability to interact with the c-Myc promoter, thereby inducing c-Myc expression.[19] Thus, AEG-1 and c-Myc provide a feedback loop promoting tumorigenesis.

c-Myc overexpression is a very common event in HCC. Genomic amplification of 8q24.1, the locus of the c-Myc gene, is a frequent event in human HCC patients, and c-Myc expression levels correlate with poor prognosis.[17] Overexpression of c-Myc in mouse models induces HCC, whereas antisense inhibition of c-Myc reverses this process.[20-24] Using a Tet-Off system, it was documented that overexpression of c-Myc induced HCC, and turning c-Myc expression off by doxycycline (Dox) treatment in tumors resulted in marked tumor reduction with induction of differentiation.[25] Removal of Dox, hence c-Myc reactivation, immediately restored neoplastic transformation. Collectively, these studies indicate that c-Myc is sufficient to induce HCC and is required to maintain the neoplastic state.

The present studies concentrated on defining how AEG-1 and c-Myc cooperate in promoting hepatocarcinogenesis because both are overexpressed in HCC. We show that hepatocyte-specific AEG-1 and c-Myc Tg mice (Alb/AEG-1/c-Myc) develop highly aggressive, metastatic HCC, either spontaneously or DEN induced, when compared to Tg mice expressing either oncogene alone. RNA-sequencing (RNA-Seq) analysis demonstrated a distinct gene signature in the double Tg mice that might confer this aggressive phenotype. These findings shed light into new mechanisms by which oncogenes cooperate in development and progression of HCC.

Materials and Methods

Mouse Models

Alb/AEG-1 and Alb/c-Myc mice, generated in B6/CBA background under albumin promoter, were described previously.[8, 24] Alb/c-Myc was a kind gift from Dr. Snorri Thorgeirsson (National Institutes of Health[NIH]/National Cancer Institute, Bethesda, MD). Heterozygote Alb/AEG-1 and Alb/c-Myc mice were crossed to obtain wild-type (WT), single Tg, and double Tg littermates. Only male mice were used for experiments. For chemical carcinogenesis, mice were given a single intraperitoneal injection of DEN (10 µg/g) at 2 weeks of age. The Virginia Commonwealth University Institutional Animal Care and Use Committee (Richmond, VA) approved the experiments, and animals were treated in ethical and humane ways.

Cell Culture

Primary mouse hepatocytes were isolated from adult mice (3-5 months old) as previously described[8] and were cultured in Williams E medium containing NaHCO3, L-glutamine, insulin (1.5 µM), and dexamethasone (0.1 µM) at 37°C and in 5% CO2. Insulin was not added when cells were cultured in basal media. Cell proliferation was analyzed by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and bromodeoxyuridine (BrdU) incorporation assays.[2, 8] Senescence was detected by γ-H2AX foci and senescence-associated β-galactosidase assays.[8] Invasion was analyzed by Matrigel invasion assays.[2] Transformation was analyzed by focus formation assays in which primary hepatocytes were plated in confluence and allowed to form foci by overcoming contact inhibition. 804G rat bladder carcinoma cells, a kind gift from Dr. Douglas E. Brash (Yale University School of Medicine, New Haven, CT), were cultured in minimum essential medium containing 2 mM of L-glutamine, 20% fetal bovine serum, 100 U/mL of penicillin, and 100 µg/mL of streptomycin at 37°C and in 5% CO2. Cell-free extracellular matrix (ECM) was prepared by removing cytoplasm and nuclei from the deposited ECM, as previously described.[26] Briefly, 804G cells were plated at 50%-70% confluence and allowed to deposit ECM for 48 hours. Cells were washed with phosphate-buffered saline (PBS), and fresh 20 mM NH4OH was added until cells were completely lysed, as assessed microscopically. Remaining ECM was washed three times with PBS. For colony formation assay (CFA), 2.5 × 105 mouse hepatocytes were plated on 804G ECM plates and cultured for 3 weeks. Colonies were counted using ImageJ software (NIH), and colonies with more than 20 pixels were enumerated.


Total RNA, extracted using Qiagen miRNAeasy mini kit (Qiagen, Valencia, CA) from livers of 3 adult mice per group, was employed for RNA-Seq. An RNA-Seq library was prepared using an Illumina TruSeq RNA sample preparation kit and sequenced on the Illumina HiSeq2000 platform. RNA-Seq libraries were pooled together to aim approximately 25-40-M read-passed filtered reads per sample. All sequencing reads were aligned with their reference genome (UCSC mouse genome build mm9; Genome Bioinformatics Group of University of California Santa Cruz, Santa Cruz, CA) using TopHat2, and the Bam files from alignment were processed using HTSeq count to obtain the counts per gene in all samples. Counts were read into R software (R Foundation for Statistical Computing, Vienna, Austria) using the DESeq package,[27] and plot distributions were analyzed using reads per kilobase per million (RPKM) values. Data were filtered based on low count or low RPKM value (<40th percentile). Pair-wise tests were performed between each group using the functions in DESeq. Genes showing absolute fold change of >2, false discovery rate of <0.1, and P value of <0.01 were selected.

Reverse-Transcription Polymerase Chain Reaction

Total RNA was used for complementary DNA (cDNA) synthesis using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA). The primers used were: Rian: sense: 5′ CT AAGTCACCATGGAGAACCAG 3′, antisense: 5′ CA GATTTCAGGGTGGCAGTAAG 3′; Mirg: sense: 5′ CCACCATCATCAGAGAGCTTC 3′, antisense: 5′ GATGTAGCCTCTGGAGTCCTT 3′; Meg3: sense: 5′CCACCATCATCAGAGAGCTTC 3′, antisense: 5′ GAGCGAGAGCCGTTCGATG 3′; Tubb5: sense: 5′ GATCGGTGCTAAGTTCTGGGA 3′, antisense: 5′ AGGGACATACTTGCCACCTGT 3′.

Statistical Analysis

Data were represented as the mean ± standard error of the mean (SEM) and analyzed for statistical significance using one-way analysis of variance, followed by Newman-Keuls' test as a post-hoc test. Additional Bayesian analysis is provided in the Supporting Information.


We checked whether overexpression of AEG-1 and c-Myc might induce spontaneous hepatocarcinogenesis and metastasis. Alb/AEG-1 mice do not develop spontaneous tumors,[8] whereas Alb/c-Myc mice develop HCC at ∼10 months of age.[22] However, distant metastases are not detected in Alb/c-Myc mice. We sacrificed one cohort of mice at 12 months of age with the presumption that, at this age, Alb/AEG-1/c-Myc mice might develop aggressive HCC with distant metastases. At this age, WT and Alb/AEG-1 mice did not develop any tumors, except for isolated nodules that were less than 2-mm in size (Table 1; Fig. 1A). Alb/c-Myc mice presented with multiple small nodules that were less than 5 mm in size. However, Alb/AEG-1/c-Myc mice developed tumors that were predominantly more than 5 mm in size (Table 1). At this time point, metastases to lungs were not detected. We analyzed a second cohort of mice at 18 months of age. At this time point, WT and Alb/AEG-1 mice developed few nodules less than 2 mm in size (Table 1; Fig. 1A). Alb/c-Myc mice developed large tumors predominantly 10-20 mm in size. However, Alb/AEG-1/c-Myc mice presented with more-aggressive phenotype with very large tumors, 10-20 mm or more than 20 mm in size, and in 3 mice the whole liver was converted into HCC with convergence of all nodules (Table 1). Most strikingly, all Alb/AEG-1/c-Myc mice developed lung metastases, which were not detected in any other groups of animals (Table 1).

Table 1. No. of Liver Nodules and Lung Metastasis Score in Spontaneously Developed HCC
  12 m 18 m
 ID No.1-2 mm2-5 mm5-10 mm10-20 mm>20 mmID No.1-2 mm2-5 mm5-10 mm10-20 mm>20 mmLung mets
WT104     17      
 108     18      
 1131    28      
 114     29      
 116     337     
 125     54      
 127     72      
 130     74      
 134     84      
AEG-1109     9      
 121     25      
 124     26      
 131     27      
 141     461     
 142     504     
 1451    58      
 154     73      
 155     76      
 158     78      
c-Myc1231 1  311  4  
 1383    32  324 
 14741   35 2 1  
 149 2   49  111 
 1506    55 6 8  
 15134   59    3 
 15213   75 5 1  
 1535    77  3   
 1567    87  11  
 15912   97 6 2  
 16045   102 3 7  
 16137   118  1 1 
       126 51 1 
       139 432  
       143  23  
       144  51  
AEG-1/c-Myc110 3 1 19  598+
 1115 6  34 14117+
 112442  47  7144+
 122  3  48  8109+
 129 1 1 61Whole liver+
 137    162  1956+
 140   2 65Whole liver+
 146  4  68Whole liver+
 148  3  94  61712+
 157   1 95 28511+
Figure 1.

AEG-1 and c-Myc overexpression markedly augments spontaneous hepatocarcinogenesis. (A) Photograph of livers of the indicated groups at 12 and 18 months. (B) Bayesian analysis to determine probabilities of no tumor and number of tumors. (C) Hematoxylin and eosin staining of liver sections at 18 months showing steatosis in Alb/AEG-1, hyperproliferative HCC in Alb/c-Myc, and HCC in a mixed background of steatosis and hyperproliferation in Alb/AEG-1/c-Myc. (D) Liver sections at 18 months from the indicated groups were stained for AEG-1, c-Myc, AFP, and PCNA. The panels show representative photomicrographs. For (C) and (D), magnification 400×.

We performed Bayesian statistical analysis to check the probability of tumorigenesis upon overexpression of AEG-1 and c-Myc using the following response variables: number and size of the nodules and the status of metastasis (Fig. 1B). The detailed methodology of analysis is presented in the Supporting Information. The probability of no tumor formation was profoundly low in Alb/c-Myc and Alb/AEG-1/c-Myc mice (Fig. 1B). However, the probability of the number of tumors was significantly higher in Alb/AEG-1/c-Myc mice, when compared to other groups (Fig. 1B). Further analysis of probabilities based on tumor sizes also showed increased probability of larger tumors in Alb/AEG-1/c-Myc mice, when compared to other groups (Supporting Figs. 1 and 2).

Histologically, WT mice maintained normal liver architecture, Alb/AEG-1 mice presented with steatosis, as described previously,[8] Alb/c-Myc mice showed features of hyperproliferative HCC,[22] and Alb/AEG-1/c-Myc mice developed HCC over a steatotic background (Fig. 1C). Liver and tumor sections at 18 months of age were stained for AEG-1, c-Myc, HCC marker α-fetoprotein (AFP), and the proliferative marker, proliferating cell nuclear antigen (PCNA; Fig. 1D). In WT liver, low-level AEG-1 was detected, which was predominantly nuclear.[8] Alb/AEG-1 livers showed increased AEG-1 expression, both in the cytoplasm and nucleus, which is consistent with the finding that overexpressed AEG-1 tends to accumulate in the cytoplasm.[8] Interestingly, AEG-1 overexpression was detected also in Alb/c-Myc tumors, where AEG-1 was detected almost exclusively in the cytoplasm, a feature observed for overexpressed AEG-1 in tumors.[2] Alb/AEG-1/c-Myc tumors showed a further increase in cytoplasmic AEG-1 staining (Fig. 1D, top row). Low-level nuclear c-Myc was detected in WT and Alb/AEG-1 liver, whereas c-Myc overexpression was detected both in the nucleus and cytoplasm in Alb/c-Myc and Alb/AEG-1/c-Myc tumors (Fig. 1D, second row). Tumors were positive for AFP (Fig. 1D, third row). Compared to WT liver, increased PCNA staining was detected both in Alb/AEG-1 liver and Alb/c-Myc tumor, the intensity of the staining being more pronounced in Alb/c-Myc tumors (Fig. 1D, bottom row). The intensity of PCNA staining was markedly augmented in Alb/AEG-1/c-Myc mice (Fig. 1D, bottom row). PCNA-positive cells per field for WT, Alb/AEG-1, Alb/c-Myc, and Alb/AEG-1/c-Myc were 0.75 ± 0.35, 12 ± 4, 14 ± 3, and 45 ± 7, respectively, with counts from nine independent fields. These findings indicate that both AEG-1 and c-Myc are capable of stimulating proliferation; however, their combinatorial effect is significantly more pronounced.

The response of these four groups of mice to DEN-induced hepatocarcinogenesis was checked next. We chose the initial time point of 28 weeks, when Alb/AEG-1 mice develop HCC whereas WT mice do not.[8] Very few isolated nodules, less than 2 mm in size, were detected in WT mice (Table 2; Fig. 2A). The number of nodules was significantly increased in all three other groups. However, the total number and size of the nodules were significantly increased in Alb/AEG-1/c-Myc mice (Table 2). More important, at 33 weeks, the whole liver of Alb/AEG-1/c-Myc mice was converted into HCC with lung metastases (Table 2; Fig. 2A). Probability analysis of these mice demonstrated that, although all groups showed very low probability of developing no tumors, the probability of having increased number of tumors was high in Alb/c-Myc mice and was markedly high in Alb/AEG-1/c-Myc mice (Fig. 2B). A similar pattern of probability was observed when size of the tumors was used as variables (Supporting Figs. 3 and 4). Histological analysis at 28 weeks showed HCC developing on a steatotic background in Alb/AEG-1 mice, Alb/c-Myc mice showed hypercellular HCC, and Alb/AEG-1/c-Myc mice presented with a mixed phenotype (Fig. 2C). The metastatic lung nodules stained positively for AFP, AEG-1, and c-Myc, confirming their hepatic origin (Fig. 2D,E).

Table 2. No. of Liver Nodules and Lung Metastasis Score in DEN-Treated Mice
  28 weeks 33 weeks
 ID No.1-2 mm2-5 mm5-10 mm10-20 mm>20 mmID No.1-2 mm2-5 mm5-10 mm10-20 mm>20 mmLung mets
WTC10     C4118 3   
 C132    C58 44   
 C161    C6110 1   
 C205    C824     
 C273    C8634 2   
 C29 2   C904212    
 C30 1   C914 1   
 C341    C93 405   
 C352    C9635 2   
       C10211 1   
AEG-1C1232 2  C79 44123  
 C2227    C89 241941 
 C5049  1 C92 30 61 
 C51 481  C95 82441 
 C653312   C97 72312 
 C8149    C106 561152 
 C8355 2  C111 48441 
c-MycC386 5 1C37 68 52 
 C945 32 C42 119731 
 C1193 5 1C57 745 2 
 C31595  2C60 67622 
 C4952 4  C98 80413 
 C5232 331C100 91733 
 C5441 4  C101 100811 
 C6352 31 C103 152342 
AEG-1/c-MycC4 56311C36Whole liver+
 C17 123111C69Whole liver+
 C21 120 34C75Whole liver+
 C28 59 32C78Whole liver+
 C45 57 5 C84Whole liver+
 C46 603 5C87Whole liver+
 C48 431053C88Whole liver+
 C55 71924C94Whole liver+
 C56 281223C99Whole liver+
 C80 84124C105Whole liver+
       C107Whole liver+
       C109Whole liver+
       C110Whole liver+
Figure 2.

DEN-induced HCC and metastases are markedly increased in Alb/AEG-1/c-Myc mice. (A) Photograph of livers of the indicated groups at 28 and 33 weeks. (B) Bayesian analysis was performed in DEN-treated 28- and 33-week-old WT, Alb/AEG-1, Alb/c-Myc, and Alb/AEG-1/c-Myc samples. (C) Hematoxylin and eosin staining of liver sections at 33 weeks. (D) Hematoxylin and eosin staining of lung sections at 33 weeks. (E) Lung tumors were stained for AFP, AEG-1, and c-Myc to demonstrate metastases from the liver. For (C-E), magnification 400×.

We analyzed primary hepatocytes from these mice to interrogate the contribution of AEG-1 and c-Myc to different hallmarks of cancer. In vitro, WT hepatocytes might replicate once before senescing ∼72-96 hours. In an 8-day assay measuring proliferation by MTT and BrdU incorporation assays, AEG-1 and c-Myc alone provided significant proliferative advantage versus WT (Fig. 3A, top and middle panels). However, the proliferative advantage provided by overexpression of both AEG-1 and c-Myc was markedly higher, compared to either oncogene alone. Culturing the hepatocytes in basal medium containing no growth factor (insulin) extended these observations. In this scenario, overexpression of AEG-1 in Alb/AEG-1 or Alb/AEG-1/c-Myc group showed an equal level of protection, which was markedly higher than that of WT, whereas overexpression of c-Myc alone provided small, albeit significant, protection, indicating that it is AEG-1 that contributes to growth-factor–independent survival (Fig. 3A, bottom panel). To obtain insights into which prosurvival pathways provide survival advantage, we analyzed the activation profile of oncogenic signal transduction pathways on days 1 and 7 of culture. Sustained activation of signaling pathways at day 7 indicates that these pathways might contribute to survival and proliferative advantages in long-term culture and hence in vivo. It should be noted that the western blottings for days 1 and 7 were performed separately, so that comparison was analyzed among the four groups in each time point, rather than between days 1 and 7. Epidermal growth factor (EGF) receptor (EGFR) activation was observed upon overexpression of either AEG-1 or c-Myc, and this activation was further augmented in Alb/AEG-1/c-Myc, which was sustained up to day 7 (Fig. 3B). Similar findings were observed for EGFR downstream-signaling molecules, protein kinase B (AKT), extracellular signal-regulated kinase (ERK)1/2, and p38 mitogen-activated protein kinase (MAPK), and for signal transducer and activator of transcription 3 (STAT3), c-Met, and c-Src (Fig. 3B). Robust up-regulation of p38 MAPK was observed on day 7 only in Alb/AEG-1/c-Myc mice. Up-regulation of antiapoptotic molecules, B-cell lymphoma 2 (Bcl-2), X-linked inhibitor of apoptosis protein (Xiap), and Survivin, but not B-cell lymphoma-extra large (Bcl-xL) and cellular inhibitor of apoptosis protein 1 (Ciap1), was observed with either oncogene alone, which was further augmented in Alb/AEG-1/c-Myc mice and sustained until day 7 (Fig. 3C).

Figure 3.

Cell proliferation and survival, as well as prosurvival signaling pathways are activated in Alb/AEG-1/c-Myc hepatocytes. (A) Top panel: Cell proliferation in standard media was determined by MTT assay. Middle panel: BrdU incorporation assay. Bottom panel: Cell proliferation in basal media without insulin was determined by MTT assay. Data represent mean ± SEM of three independent experiments. *P < 0.01 versus WT; #P < 0.05 versus WT. (B) Western blotting analysis of the indicated proteins in cell survival pathways. (C) Western blotting analysis of the indicated antiapoptotic proteins. For (B) and (C), for each blotting, EF1α was used for loading control and one representative EF1α blot is shown here. EF1α, elongation factor 1 alpha.

We previously demonstrated that overexpression of AEG-1 protects hepatocytes from senescence.[8] Analysis of senescence by γ-H2AX foci assay demonstrated profound induction of senescence in WT hepatocytes by day 7 of culture (Fig. 4). Very few isolated γ-H2AX foci were detected in Alb/AEG-1 and Alb/AEG-1/c-Myc hepatocytes, whereas a few foci were observed in Alb/c-Myc hepatocytes. These findings were also confirmed by senescence-associated β-galactosidase assay (Supporting Fig. 5). These findings reflect c-Myc-induced senescence, as described before,[28] which might be overcome by overexpression of AEG-1. Collectively, activation of prosurvival and proliferative pathways, up-regulation of antiapoptotic proteins, and inhibition of senescence contribute to the aggressive behavior observed in Alb/AEG-1/c-Myc mice.

Figure 4.

Senescence is inhibited in Alb/AEG-1/c-Myc hepatocytes. Hepatocytes from the indicated groups were stained with DAPI for nucleus and for γ-H2AX at days 1, 4, and 7 of culture. The images were analyzed by a confocal laser scanning microscope. Magnification, 630×. DAPI, 4',6-diamidino-2-phenylindole.

We next analyzed transformation by focus formation assay using primary hepatocytes. c-Myc expression alone provided a greater transforming advantage over AEG-1; however, the combination of the two oncogenes augmented transformation further (Fig. 5A). As a corollary, we performed CFAs for 3 weeks, in which the hepatocytes were plated on an endogenous matrix, which was created by first plating 804G cells that were then gently treated with alkaline solution to remove the cells, leaving behind the matrix proteins. WT hepatocytes formed no colonies, whereas Alb/AEG-1 and Alb/c-Myc hepatocytes formed small colonies. However, the colonies formed by Alb/AEG-1/c-Myc hepatocytes were markedly increased in number as well as in size (Fig. 5B). In Matrigel invasion assays, c-Myc overexpression conferred a more invasive advantage over AEG-1, and together AEG-1 and c-Myc increased invasion further (Fig. 5C). To obtain insights into the molecular mechanism of these phenomena, we checked expression profiles of key players regulating epithelial-mesenchymal transition (EMT; Fig. 5D). The changes in EMT regulators and markers were more pronounced on day 7 of culture. E-cadherin level was decreased only in Alb/c-Myc mice, but not in the other three groups (Fig. 5D). It was documented previously that E-cadherin is down-regulated in livers of only c-Myc Tg mice, but not in c-Myc/TGFα or c-Myc/E2F1 double Tg mice.[29] Increased N-cadherin levels were observed only in Alb/AEG-1/c-Myc on day 7. A decrease in zona occludens 1 (ZO-1) and claudin-1, and increases in transcription factor 8 (TCF8), Snail, Slug, vimentin, and secreted protein acidic and rich in cysteine (SPARC) were observed in the Tgs, when compared to WT mice. Vimentin and SPARC expression could not be detected on day 1 of culture. A robust increase in the EMT-inducing transcription factors, Snail and Slug, was observed only in Alb/AEG-1/c-Myc at day 7. Additionally, activation of β-catenin, and transforming growth factor beta (TGF-β) and its downstream signaling, indicated by activation of mothers against decapentaplegic (Smad)-2 and Smad-3, was observed for Alb/AEG-1/c-Myc, especially at day 7. These findings suggest that Alb/AEG-1/c-Myc maintains a sustained EMT phenotype that might contribute to invasion and metastasis.

Figure 5.

Transformation and invasion are augmented in Alb/AEG-1/c-Myc hepatocytes. (A) Top panel: photomicrographs of the cells. Bottom panel: graphical representation of quantification of the number of foci. (B) Top panel: photographs of the plates showing the colonies. Bottom panel: graphical representation of quantification of the number of colonies. (C) Matrigel invasion assay. Top panel: photomicrographs of the invading cells. Bottom panel: graphical representation of the quantification of the number of invading cells. For (A-C), data represent mean ± SEM of three independent experiments. *P < 0.01 versus WT; #P < 0.05 versus Alb/AEG-1 or Alb/c-Myc. (D) Western blotting analysis of the indicated proteins regulating EMT. For each blotting, EF1α was used for loading control and one representative EF1α blotting is shown here. EF1α, elongation factor 1 alpha.

To obtain insights into the molecular mechanism by which AEG-1 and c-Myc cooperate to promote the aggressive hepatocarcinogenic phenotype, we performed RNA-Seq analysis using liver samples from naïve adult WT, Alb/AEG-1, Alb/c-Myc, and Alb/AEG-1/c-Myc mice. With an absolute fold change of >2-fold, 30 and 35 genes showed differential change in Alb/AEG-1 and Alb/c-Myc, respectively, versus WT. However, a staggering number of 258 genes showed differential change in Alb/AEG-1/c-Myc versus WT (Fig. 6A; Supporting Table 1). Of these 258 genes, 121 showed up-regulation and 137 showed down-regulation (Supporting Table 1). Whereas some genes showed common changes in single and double Tgs, when compared to WT, Alb/AEG-1/c-Myc mice showed exclusive and robust change in expression of 71 genes (Supporting Fig. 6A). We performed biological processes and pathway analysis using the software DAVID (Database for Annotation, Visualization and Integrated Discovery; Among the biological processes differentially regulated in Alb/AEG-1/c-Myc mice, two major classes were prominent: (1) cell cycle and cell division and (2) metabolic processes, including lipid metabolic processes (Supporting Table 2). Pathway analysis showed a differential change in MAPK-signaling pathways, metabolic signaling, and complement and coagulation cascade pathways in Alb/AEG-1/c-Myc versus WT (Supporting Table 3). We also performed network analysis using Ingenuity software. The top three networks of genes that showed significant changes in Alb/AEG-1/c-Myc versus WT include (1) cancer, cellular function, and maintenance, (2) lipid metabolism, and (3) cell death and survival (Supporting Fig. 6B). These networks and pathway analyses included gene ontology that is already well known and genes regulated by c-Myc and AEG-1, as described in previous studies.[8, 24, 30, 31] Additionally, changes in many of these genes in Alb/AEG-1/c-Myc liver have also been observed in human HCC (a review of the literature is provided in Supporting Table 4). Very interestingly, we observed robust up-regulation of several noncoding RNAs (ncRNAs)—Rian, Mirg, and Meg3—only in the Alb/AEG-1/c-Myc group (Supporting Table 1). Overexpression of these three ncRNAs was confirmed by reverse-transcriptase polymerase chain reaction (RT-PCR; Fig. 6B and Supporting Fig. 7A). We next attempted to knock down these three ncRNAs by four different small interfering RNAs (siRNAs) for Rian and Meg3 and three different siRNAs for Mirg (Supporting Table 5). For Rian and Mirg, two independent siRNAs effectively knocked down the corresponding RNA by >50% (Supporting Fig. 7B). However, none of the four siRNAs worked for Meg3. Knocking down Rian and Mirg significantly inhibited proliferation of Alb/AEG-1/c-Myc hepatocytes, and the combination of these two siRNAs inhibited proliferation further (Fig. 6C, left and middle panels). Rian and Mirg siRNA inhibited invasion by Alb/AEG-1/c-Myc hepatocytes (Fig. 6C, right panel). However, the combination did not inhibit invasion further, suggesting that these two ncRNAs might regulate the same molecules mediating invasion. Knocking down Rian or Mirg resulted in a substantial decrease in proliferation markers survivin and PCNA, inhibition of activation of p38 MAPK, and down-regulation of EMT markers vimentin and TCF8 (Fig. 6D). There was a small decrease in Snail, but no decrease in Slug was observed. These findings suggest that these ncRNAs, induced by AEG-1 and c-Myc, contribute both to proliferative and invasive advantages.

Figure 6.

AEG-1 and c-Myc overexpression specifically induces ncRNAs. (A) Heat map showing differential gene expression in the indicated groups. (B) RT-PCR analysis for Rian, Mirg, and Meg3 in the indicated groups. Tubb5 was used as a control. Asterisk indicates nonspecific band. (C) Analysis of cell proliferation and invasion upon knockdown of Rian and Mirg. Left panel: Cell proliferation by MTT assay at 72 hours after transfection. Middle panel: BrdU incorporation assay at 72 hours after transfection. Right panel: Matrigel invasion assay at 48 hours after transfection. Rian and Mirg ncRNAs were knocked down by two individual siRNAs either alone or in combination. NTC si, control scrambled siRNA. Data represent mean ± SEM of three independent experiments. *P < 0.05 versus WT; **P < 0.01 versus No si or NTC si. (D) Western blotting analysis of the indicated proteins. For each blotting, EF1α was used for loading control and one representative EF1α blotting is shown here. EF1α, elongation factor 1 alpha.


We document that overexpression of AEG-1 and c-Myc resulted in spontaneous HCC with lung metastases, which takes more than 12 months to develop. Cancer genome sequencing indicates that most cancers require at least five genetic changes to evolve.[32] Thus, overexpression of two oncogenes might not be enough for generation of early aggressive disease. Therefore, initiation of additional mutations by DEN treatment significantly accelerated the aggressive behavior conferred by AEG-1 and c-Myc. RNA-Seq analysis identified unique changes in gene expression only when AEG-1 and c-Myc were overexpressed. The predominant biological processes that were modulated in Alb/AEG-1/c-Myc liver versus WT liver involve cell cycle and cell division and metabolic pathways, including lipid metabolism. These findings are expected, considering the pivotal role of c-Myc in regulating cell cycle and cell division and cellular metabolic processes.[24, 33-35] Alb/AEG-1 mice develop steatosis, and in this context, alteration in lipid-metabolism–associated genes is also expected.[8] Pathways analysis identified changes in the MAPK-signaling pathway, which has been documented for AEG-1, metabolic pathways, and complement and coagulation cascade pathways. Indeed, our previous study aimed at characterizing Alb/AEG-1 mouse unraveled a key role of coagulation factors in mediating AEG-1-induced angiogenesis.[8] Therefore, the gene expression changes that were observed in Alb/AEG-1/c-Myc mice reflected gene expression changes of either oncogene alone. It should be noted that we used a stringent gene expression change of >2-fold, which limited the genes changed in either Alb/AEG-1 or Alb/c-Myc to a small number. In Alb/AEG-1/c-Myc mice, these changes became more pronounced, thereby significantly increasing the total number of genes with detectable alterations. These observations suggest that AEG-1 and c-Myc might cooperate to amplify expression of genes regulated by either oncogene alone. This cooperation might be conferred by reciprocal positive regulatory changes, for example, AEG-1 induces c-Myc by activating the Wnt/β-catenin pathway and also by squelching away a c-Myc-specific transcriptional repressor, PLZF.[2, 19] On the other hand c-Myc itself transcriptionally regulates AEG-1. Whether AEG-1 facilitates transcriptional activity of c-Myc remains to be determined. Although AEG-1 does not bind to DNA, it interacts with the p65 subunit of nuclear factor kappa B (NF-κB) and CREB-binding protein, thereby functioning as a bridging factor between NF-κB and basal transcriptional machinery promoting transcriptional activity of NF-κB.[36] Whether a similar relationship exists between AEG-1 and c-Myc needs to be explored.

However, one very striking observation is the induction of ncRNAs as a consequence of overexpression of both AEG-1 and c-Myc. The delta-like 1/deiodinase, iodothyronine, type III (DLK1-DIO3) genomic region, located on mouse chromosome 12 and on human chromosome 14, codes for genes that are expressed either maternally or paternally.[37-40] Rian, Meg3, and Mirg are maternally expressed genes that generate ncRNAs.[38] Within these ncRNAs lie several small nucleolar RNAs and miRNAs, which have been recognized to regulate tumorigenesis in diverse organs.[41] Targeted overexpression of Rian and Mirg resulted in spontaneous HCC development in mice.[42] The miRNA cluster generated from the DLK1-DIO3 region has been implicated to confer a stem-like subtype in human HCC associated with poor survival.[43, 44] The overexpression of these ncRNAs in Alb/AEG-1/c-Myc might mediate an aggressive oncogenic phenotype in these mice, which is supported by the in vitro analysis of proliferation and invasion upon knockdown to these ncRNAs. It is intriguing to observe that overexpression of both AEG-1 and c-Myc is required for expression of these imprinted genes.

In vitro analysis demonstrated that although overexpression of both AEG-1 and c-Myc provided proliferative advantage to hepatocytes, it is AEG-1 that enabled the cells to survive in a harsh environment, such as growth factor deprivation. On the contrary, AEG-1 alone did not confer transforming ability to hepatocytes, but contributed to transformation induced by c-Myc overexpression. These findings suggest that c-Myc overexpression is able to transform the cells, whereas AEG-1 overexpression allows these transformed cells to progress further with increasing invasive and metastatic abilities. Compared to Alb/AEG-1 and Alb/c-Myc, Alb/AEG-1/c-Myc showed robust, sustained activation of all prosurvival signaling pathways and multiple EMT modulators. Most important, EGF-, TGF-β-, and β-catenin-signaling pathways, which are key determinants of all hallmarks of aggressive HCC,[45] showed sustained activation in Alb/AEG-1/c-Myc mice. Thus, a combination of gene expression changes and activation of multiple oncogenic signal transduction pathways contribute to the aggressive behavior conferred by AEG-1 and c-Myc.

In summary, we have generated a novel mouse model of aggressive HCC. Alb/AEG-1/c-Myc mice provide a valuable animal model to inhibit tumor-suppressor proteins to analyze consequences of oncogene activation and tumor-suppressor mutations during hepatocarcinogenesis. Additionally, this mouse model might be effectively used for monitoring efficacy of novel, targeted therapies.

Author names in bold designate shared co-first authorship.