SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Discussion
  5. Acknowledgment
  6. References
  7. Supporting Information

Studies on gene and/or microRNA (miRNA) dysregulation in the early stages of hepatocarcinogenesis are hampered by the difficulty of diagnosing early lesions in humans. Experimental models recapitulating human hepatocellular carcinoma (HCC) are then used to perform this analysis. We performed miRNA and gene expression profiling to characterize the molecular events involved in the multistep process of hepatocarcinogenesis in the resistant-hepatocyte rat model. A high percentage of dysregulated miRNAs/genes in HCC were similarly altered in early preneoplastic lesions positive for the stem/progenitor cell marker cytokeratin-19, indicating that several HCC-associated alterations occur from the very beginning of the carcinogenic process. Our analysis also identified miRNA/gene-target networks aberrantly activated at the initial stage of hepatocarcinogenesis. Activation of the nuclear factor erythroid related factor 2 (NRF2) pathway and up-regulation of the miR-200 family were among the most prominent changes. The relevance of these alterations in the development of HCC was confirmed by the observation that NRF2 silencing impaired while miR-200a overexpression promoted HCC cell proliferation in vitro. Moreover, T3-induced in vivo inhibition of the NRF2 pathway accompanied the regression of cytokeratin-19-positive nodules, suggesting that activation of this transcription factor contributes to the onset and progression of preneoplastic lesions towards malignancy. The finding that 78% of genes and 57% of dysregulated miRNAs in rat HCC have been previously associated with human HCC as well underlines the translational value of our results. Conclusion: This study indicates that most of the molecular changes found in HCC occur in the very early stages of hepatocarcinogenesis. Among these, the NRF2 pathway plays a relevant role and may represent a new therapeutic target. (Hepatology 2014;58:228–241)

Abbreviations
2-AAF

2-acetylaminofluorene

aHCC

advanced HCC

BH

Benjamini-Hochber

DENA

diethylnitrosamine

eHCC

early HCC

GCLC

glutamate-cysteine ligase

GST

glutathione S-transferase

GSTP

placental glutathione S-transferase

IHC

immunohistochemistry

IL-1

interleukin-1

IPA

Ingenuity Pathway Analysis

KEAP1

kelch-like ECH-associated protein 1

KRT-19

cytokeratin-19

LPS

lipopolysaccharide

MAF

musculoaponeurotic fibrosarcoma oncogene homolog

miRNA

microRNA

NQO1

NAD(P)H dehydrogenase, quinone 1

NRF2

nuclear factor erythroid related factor 2

qRT-PCR

quantitative reverse transcriptase polymerase chain reaction

R-H model

Resistant-Hepatocyte model

RXR

retinoic-X-receptor

T3

thyroid hormone

Hepatocellular carcinoma (HCC) is the third cause of cancer-related deaths worldwide and a major health problem. Liver cirrhosis is the underlying disease in more than 80% of cases and can be due to different etiologies such as hepatitis B and C, and nonalcoholic and alcoholic fatty liver disease.[1]

Although many factors, including chromosomal anomalies, genetic polymorphisms, genetic and epigenetic alterations contribute to HCC onset and progression,[2] the relevant molecular mechanisms remain largely unclear. Recently, genome-wide gene expression microarray and quantitative reverse-transcription polymerase chain reaction (qRT-PCR) studies have tried to identify the genes abnormally expressed in HCC and to generate molecular signatures able to classify different types of liver tumors and distinguish progressive from nonprogressive lesions.[3] These studies indicate a general aberrant activation of signaling pathways involved in cellular proliferation, survival, differentiation, and angiogenesis, but a general consensus for a signature or even a single pathway prominent and characteristic of this cancer is still missing.

Recent studies have shown that classification of tumors can also be performed by evaluating the modulation of microRNAs (miRNAs), small noncoding RNAs acting as regulators of gene expression.[4] MiRNAs are dysregulated in several cancers where they can behave as oncogenes[5] or tumor suppressor genes,[6] depending on the cellular function of their targets. As far as HCC is concerned, several miRNAs have been consistently found up- or down-regulated in different study cohorts.[7, 8] Moreover, a putative role of miRNAs in predicting tumor outcome has been assessed on the basis of their expression.[9] However, although miRNA changes between tumor specimens and the normal corresponding tissues have been investigated in HCC, no clear-cut conclusion has been reached from most of the various microarray-based studies. It is now becoming increasingly evident that the molecular pathogenesis of HCC cannot be understood without more detailed studies of the molecular alterations characterizing its early development. Indeed, by investigating fully developed HCC it is difficult to discriminate changes causing cancer progression from those being the consequence of cell transformation. Comprehensive studies on gene expression patterns and/or miRNA dysregulation in the early stages of liver carcinogenesis are, so far, scanty and the clinical and diagnostic value of gene expression changes occurring in human HCC remains elusive. Thus, investigations on the precancerous lesions and early stage liver cancers are likely to reveal useful insights into the molecular mechanisms underlying the multistep process of hepatocarcinogenesis.

Since studies on initial HCC stages in humans are hampered by the clinical difficulty of diagnosing early lesions, experimental models allowing us to dissect the several steps of HCC are mandatory. Among the animal models used to characterize the process of HCC development, the Resistant-Hepatocyte (R-H) model allows dissecting the different steps of the carcinogenic process, offering the possibility to identify phenotypically distinct lesions at well-defined timings.[10] Histologically, livers sequentially develop preneoplastic nodules (1-2 months after initiation with diethylnitrosamine [DENA]) arising in a fibrotic environment, adenomas and early HCCs (Ade and eHCC, 10 months) and fully advanced HCCs (aHCC, 14 months). A translational value of this model has been recently shown, since the gene expression signature associated with the rat lesions (positive for the stem/progenitor cell marker cytokeratin-19 [KRT-19]) can successfully predict the clinical outcome of human HCC.[11] The finding that the KRT-19+ HCC subtype is characterized by the worst clinical prognosis among all human HCC subclasses[12] suggests that KRT-19 is a potential prognostic marker for HCC.

The aim of the present study was to perform an integrative analysis of global miRNA and messenger RNA (mRNA) expression profiles in the R-H model of hepatocarcinogenesis for enhanced marker and therapeutic target discovery. Specifically, we aimed at (1) identifying miRNAs/genes dysregulated during the carcinogenic cascade, mainly focusing on the less-investigated early steps; (2) analyzing miRNA/mRNA correlations to unveil integrated networks that are altered at the beginning of the process and maintained along tumor progression; and (3) validating the translational value of this rat model also for miRNA studies, by conducting comparative analyses between miRNAs and mRNAs dysregulated in rat preneoplastic and neoplastic lesions and those identified in human HCCs.

We demonstrate here that several deregulated miRNAs/genes in fully developed rat HCC, including many miRNAs/genes altered in human HCC, are already dysregulated in the very early step of tumorigenesis. Importantly, our findings unveil the activation of the nuclear factor erythroid related factor 2 (NRF2) transcription factor pathway from the very beginning and throughout the process and they also reveal the existence of regulatory networks between miRNAs and their target genes. In particular, we found up-regulation of miR-200a that controls the NRF2 pathway. Finally, we show that a high number of dysregulated miRNAs/genes in rat preneoplastic and neoplastic lesions are dysregulated in primary human HCC as well, suggesting the potential utility of this model to investigate into the critical molecular changes underlying HCC development.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Discussion
  5. Acknowledgment
  6. References
  7. Supporting Information

Animals and Treatment

Guidelines for Care and Use of Laboratory Animals were followed during the investigation. All animal procedures were approved by the Ethical Commission of the University of Cagliari and the Italian Ministry of Health. Male Fischer F-344 rats (100-125 g) were purchased from Charles River (Wilmington, MA). Preneoplastic lesions and HCCs were induced as described in the Supporting Material. Histologic classification of preneoplastic nodules, adenomas, eHCCs, and aHCCs was performed as described.[13]

MiRNA and mRNA Expression Profiling

RNA was extracted and purified from each individual lesion after laser microdissection from the liver of four to five animals (for microdissection procedures, see Supporting Material). The miRNA expression profiling was performed using TaqMan Low Density Arrays starting from 600 ng of RNA and following the manufacturer's instructions (Applied Biosystems, Carlsbad, CA). For the gene expression profile, 150 ng of RNA were amplified (Illumina TotalPrep RNA Amplification Kit), labeled and hybridized on Illumina microarray (RatRef-12 V1 BeadChips, Illumina, San Diego, CA), including 21,791 gene-specific oligonucleotide probes (for further details and data analysis, see Supporting Material). Gene array data are available at GEO, accession number GSE44106. MiRNAs and mRNAs validation was performed using specific TaqMan assays (Applied Biosystems). For further information on Materials and Methods see Supporting Material.

Results

MiRNome Classifies the Different Steps of Hepatocarcinogenesis and Identifies miRNA Changes Occurring in Very Early Phases of the Tumorigenic Process

To generate miRNA expression signatures specific for the different steps of hepatocarcinogenesis, we analyzed microdissected nodules (10 weeks after initiation with DENA), adenomas and eHCCs (10 months) and aHCCs (14 months). Immunohistochemical analysis showed that less than 25% of preneoplastic nodules, equally positive for the placental form of placental glutathione S-transferase (GSTP), were also positive for KRT-19 (Supporting Table 1). Notably, almost all aHCCs examined at the end of the experiment showed positivity for KRT-19, supporting our preliminary findings that, in this model, KRT-19-positive preneoplastic lesions are the progenitors of HCC, while those negative for KRT-19 are more likely to spontaneously regress.[11]

In microdissected lesions, 200 out of 375 analyzed miRNAs were detectable by TaqMan low-density array technology and were further considered in the present study (see Supporting Material for selection criteria). Unsupervised hierarchical clustering, based on the relative expression levels of the different miRNAs, revealed the existence of two major clusters, separating early preneoplastic lesions from more advanced stages (Fig. 1A). Within the two clusters, the miRNome was able to classify the different types of lesions; moreover, a more stringent selection of differentially expressed miRNAs allowed a nearly complete separation also between concomitant adenomas and early carcinomas (Supporting Fig. 1). Quantitative RT-PCR validation performed on 10 randomly selected miRNAs in individual lesions confirmed the TaqMan array results in 80% of cases (Supporting Fig. 2).

image

Figure 1. MiRNA profiling classifies the different steps of hepatocarcinogenesis. (A) Hierarchical clustering of miRNA expression profile in KRT-19- and KRT-19+ preneoplastic lesions, adenomas (Ade), early HCCs (eHCCs), and advanced HCCs (aHCCs) in rats exposed to the R-H model. Each row represents the expression profile of a miRNA expressed as ΔΔct calculated with respect to age-matched controls (see Supporting Material for details) and each column represents a sample. Different types of lesions are indicated by colored bars. Red and green colors represent higher or lower expression levels of the miRNA (median centered), respectively. (B) Number of differentially expressed miRNAs in each step compared to age-matched controls, defined by P < 0.05 (BH-corrected). (C) Venn diagram illustrating the overlapping subsets of significantly dysregulated miRNAs in HCC progression.

Download figure to PowerPoint

In order to identify miRNAs differentially expressed at each stage compared to the matched normal controls, we applied the Limma analysis package[14] (P < 0.05; Benjamini-Hochberg [BH]-corrected) (Fig. 1B). Comparing each step with the previous one, we found both miRNAs dysregulated in specific transitions (Fig. 1C; Supporting Table 2) and miRNAs commonly altered in consecutive steps (Fig. 1C; Supporting Table 3A-C). Interestingly, 13 miRNAs already modified in KRT-19+ early lesions (e.g., miR-224, miR-122, and miR-375) were altered throughout the entire process (Fig. 1C; Supporting Table 3D), suggesting their essential role in cancer development. Further support for the relevance of these miRNAs in liver carcinogenesis comes from the observation that 7 out of 13 miRNAs have been found similarly dysregulated in human HCC as well.[15]

MiRNA expression analysis between the KRT-19- and KRT-19+ preneoplastic lesions revealed that no miRNA reached the BH correction for multiple testing at the selected threshold of P < 0.05; however, using t-test for each single miRNA, we found 23 differentially expressed miRNAs (P < 0.05; Supporting Table 4) that are likely important in the progression of KRT-19+ lesions towards malignancy.

mRNA Profiling Reveals That Most of the HCC Associated Molecular Changes Already Occur in Preneoplastic Lesions

Gene expression profiling was performed in the same lesions using the Illumina microarray. A total of 1,144 out of 21,791 genes included in the array were selected as described in the Supporting Material. Hierarchical cluster analysis stratified the rat lesions into two major clusters: (1) normal liver and preneoplastic KRT-19- lesions; (2) preneoplastic KRT-19+ lesions, adenomas/eHCCs and aHCCs, forming three distinct subclusters (Fig. 2A). A major difference between transcriptome and miRNome was the ability of the latter to clearly separate preneoplastic from neoplastic lesions, still maintaining the difference between KRT-19- and KRT-19+ nodules. Quantitative RT-PCR validation performed on randomly selected genes confirmed the microarray expression data for all the examined genes (Supporting Fig. 3).

image

Figure 2. Gene profiling classifies the different steps of hepatocarcinogenesis. (A) Hierarchical clustering of 1,144 genes in normal liver (ctr), KRT-19- and KRT-19+ preneoplastic lesions, adenomas (Ade), early HCCs (eHCCs), and advanced HCCs (aHCCs) in rats exposed to the R-H model. Each row represents the expression profile of a gene and each column represents a sample. Controls and different types of lesions are indicated by colored bars. Red and green colors represent higher or lower expression levels of the mRNA (median-centered), respectively. (B) Venn diagrams illustrating overlapping subsets of significantly dysregulated genes in KRT-19- and KRT-19+ preneoplastic nodules (left) and in HCC progression (right). The differentially expressed genes between each sample and its age-matched control were defined by P < 0.01 (BH-corrected) and fold-change >2 or <-2.

Download figure to PowerPoint

To identify the differentially expressed genes in each type of lesion towards its age-matched control we applied the Limma analysis package. As shown in the Venn diagram (Fig. 2B, left), although KRT-19- and KRT-19+ lesions are histologically very similar, they exhibited a strikingly different number of modified genes. Besides the 64 dysregulated genes shared between KRT-19- and KRT-19+ likely involved in nodule formation, 602 genes were exclusively altered in KRT-19+ preneoplastic lesions, suggesting that they are relevant for nodule progression. Interestingly, 216 out of 234 altered genes in aHCC were altered in KRT-19+ nodules as well (Fig. 2B, right). Among these, 33/39 of the most up-regulated (fold change versus controls >5) and 12/15 of the most down-regulated (fold change versus controls <-5) genes in aHCC were the most dysregulated in KRT-19+ nodules as well (Table 1). These results suggest that the major expression changes leading to HCC occur in the very first stages of tumor progression.

Table 1. Most Dysregulated Genes in aHCC
GeneFC aHCC/ctrGeneFC aHCC/ctr
  1. Genes in bold are dysregulated in KRT-19+ preneoplastic lesions as well. NRF2 target genes are in italic.

Most up-regulated genesCcnb26.39
Akr1b869.80Npdc16.34
Defb159.73Bzrp6.12
Gstp151.91Pir5.93
Gstp246.60Gclc5.91
Gpx230.62Anxa15.76
Ca228.32Ica15.74
Yc225.61Esm15.54
Nqo120.54Phgdh5.26
Ddit4l19.51Traf4af15.16
Akr7a318.40Txnrd15.07
Krt1-1917.09Serpinb1a5.00
Pcp413.79Most down-regulated genes
Cryab12.34Cyp2c−69.57
Abcc311.53Obp3−49.07
Fabp411.25Ca3−29.79
Slc25a410.78Dhrs7−22.80
Anxa210.05Hao2−10.86
Tubb69.25Cdh17−10.80
Lbp9.08Olr59−9.86
S100a119.07Avpr1a−8.75
Abcb17.75Sult1c1−8.55
Scd27.54Aox3−8.47
Ugt1a67.12Cyp3a3−7.03
Ltb4dh7.09Ust5r−6.56
Cdc2a6.96Cyp2c37−5.81
Cd246.77Akr1c18−5.50
Rhbg6.42Gnmt−5.42
Ingenuity Pathway Analysis Unveils NRF2 Activation in Preneoplastic and Neoplastic Lesions

Ingenuity Pathway Analysis (IPA) of genes altered at the final stage of the carcinogenic process (aHCC) revealed that most of the dysregulated genes are involved in metabolic pathways. Among these, there are NRF2-mediated oxidative stress response, lipopolysaccharide (LPS) / interleukin (IL)-1-mediated inhibition of retinoic-X-receptor (RXR) function, aryl hydrocarbon receptor signaling, and xenobiotic metabolism (Fig. 3A). Strikingly, most of the altered pathways in aHCC were already modified in KRT-19+ preneoplastic lesions. Functional investigation also underlined common pathway modifications between early and late stages of hepatocarcinogenesis (Fig. 3B).

image

Figure 3. Functional analysis of differentially expressed genes in the progressive steps of hepatocarcinogenesis. Top five enriched pathways (A) and functions (B) in early KRT-19- and KRT-19+ preneoplastic lesions and fully developed HCC. P values were determined using the Ingenuity scoring system and were judged significant at P < 0.05 (vertical black line). (C) Differentially activated transcription factor-dependent pathways (according to IPA) in the lesions indicated in (A,B). TF: transcription factor. (D) Network visualization of the transcription factor NRF2 pathway. TF → target gene. Red: up-regulated genes; green: down-regulated genes. Additional networks and symbol legends are shown in Supporting Fig. 4.

Download figure to PowerPoint

Analysis of transcription factor-dependent pathways differentially activated in early lesions or in aHCC identified MYC and JUN/FOS (Fig. 3C; Supporting Fig. 4), already described to be involved in HCC.[16] Remarkably, among the most dysregulated there was the NRF2-dependent pathway (Fig. 3C), whose role in cancer development has recently become a topic of an important controversy.[17] The finding that the majority of NRF2 target genes were up-regulated indicates activation of the NRF2 pathway (Fig. 3D). Notably, several of these genes were among the most up-regulated in both KRT-19+ lesions and aHCCs (see Table 1). Moreover, the small musculoaponeurotic fibrosarcoma oncogene homolog (MAF) family of transcription factors, which operate as coactivators of NRF2,[18] was found activated at all stages of HCC development (Fig. 3C; Supporting Fig. 4).

Integrated miRNA:mRNA Analysis Uncovers Differentially Expressed miRNA/Gene Modules in Early Preneoplastic Lesions

To identify a possible correlation between miRNAs and genes dysregulated at the initial stage of the process, we selected conserved putative miRNA targets in rat, predicted by the TargetScanS algorithm. For each of the differentially expressed genes in KRT-19+ nodules, we extracted the annotated regulating miRNAs and performed an intersection with the differentially expressed miRNAs in the same stage. We found that a consistent percentage of modified genes are targets of dysregulated miRNAs (Fig. 4). To study the relative expression of miRNAs and of their target genes, we evaluated the number of positive and negative correlations between the predicted miRNA-target gene pairs by means of the fold-change value reported by the Limma package. The results show that 171 out of 215 miRNA/mRNA pairs (79%) shown in Fig. 4 displayed anticorrelated expression, while 44 were positively correlated.

image

Figure 4. MiRNA:mRNA integrated network in the early step of hepatocarcinogenesis. Modules represent miRNAs (orange diamonds) and genes (light blue squares). Edges connect miRNAs and putative target genes predicted by TargetScanS, both differentially expressed in KRT-19+ nodules versus controls. Module size reflects the degree of connectivity between miRNAs and target genes in the network.

Download figure to PowerPoint

In the transition from normal liver to KRT-19+ nodules we observed at least two relevant nodes, where two miRNA families, miR-30 and miR-200, control the expression of many modified genes. While expression of the miR-30 family was mainly modified at this initial stage, miR-200 family was up-regulated in both preneoplastic lesions and aHCCs (Supporting Fig. 5). Notably, one of the target genes of miR-200a is kelch-like ECH-associated protein 1 (KEAP1), a negative regulator of NRF2.[19]

NRF2 Pathway Promotes HCC Cell Growth and Is Controlled by miR-200a

NRF2 is an integrated redox sensitive signaling system that regulates 1%-10% of human genes and is negatively controlled by KEAP1, which promotes NRF2 proteasome-mediated degradation.[19] As mentioned above, the NRF2 pathway was already activated in early preneoplastic lesions and along tumor progression (Fig. 3C,D). Accordingly, immunohistochemistry (IHC) performed on lesions at different stages revealed that all of them were strongly positive for NRF2 (Fig. 5A); most important, IHC showed NRF2 nuclear staining in several preneoplastic and neoplastic hepatocytes, clearly demonstrating translocation and suggesting activation of this transcription factor. Activation of the NRF2-dependent pathway was confirmed by the increased expression of three examined NRF2 target genes[20] (NAD(P)H dehydrogenase, quinone 1 [NQO1], GSTA4, and glutamate-cysteine ligase [GCLC]), at all stages of hepatocarcinogenesis (Fig. 5B). In parallel with NRF2 nuclear translocation, a decrease of KEAP1 was observed (Fig. 5A).

image

Figure 5. NRF2 and KEAP1 expression in the different stages of hepatocarcinogenesis. (A) NRF2 (left panels) and KEAP1 (right panels) expression analysis performed by IHC in preneoplastic lesions, eHCC and aHCC. The insets show staining of nodules and normal surrounding tissue at lower magnification. Each panel is representative of five samples for each stage of the carcinogenic process (magnification ×20; inset magnification ×4). (B) qRT-PCR analysis of NRF2 target gene expression in control livers, KRT-19+ preneoplastic lesions, eHCC, and aHCC. Values are shown as fold change difference compared to normal liver (control). The histogram represents mean values ± SE of 4 to 6 animals per group *P < 0.05; **P < 0.01.

Download figure to PowerPoint

To mechanistically investigate the role of this pathway, we moved to in vitro experiments and evaluated cell growth upon NRF2 or KEAP1 silencing in three human HCC cell lines. KEAP1 silencing was associated with increased NRF2 protein levels (Fig. 6C; Supporting Fig. 6C). When cell growth was assessed in the absence or in the presence of oxidative stress (H2O2), we found that NFR2 silencing impaired cell proliferation, while KEAP1 silencing increased growth rate (Fig. 6A; Supporting Fig. 6A). Analysis of NRF2 target gene expression demonstrated either inhibition or activation of the pathway upon NRF2 or KEAP1 silencing, respectively (Fig. 6B; Supporting Fig. 6B).

image

Figure 6. Modulation of the NRF2 pathway affects HCC cell growth. (A) Mahlavu and HuH7 human HCC cells were transiently transfected with NRF2 or KEAP1 or control (siC) siRNAs or miR-200a mimic and seeded in 96-well plates. Cell growth was evaluated by crystal violet staining 72 hours after seeding, in either the presence (HuH7) or the absence (Mahlavu and HuH7 NT) of 25 μM hydrogen peroxide (H2O2). (B) Expression of NRF2 target genes NQO1, GCLC, and GSTA4 was evaluated by qRT-PCR on RNA extracted from the cells transfected in (A). Values are shown as fold change difference compared to control transfected cells (siC). (C) Cells were transiently transfected with the indicated siRNAs or miR-200a mimic. Total cell lysates were extracted 48 hours upon transfection and analyzed by western blot. β-actin was used as loading control. Quantification of KEAP1 related to the corresponding β-actin is shown for miR-200a (control siC was set as 1).

Download figure to PowerPoint

Since in humans KEAP1 is negatively controlled by miR-200a,[21] which is up-regulated in KRT-19+ lesions and aHCC compared to normal liver (Supporting Table 2, Supporting Fig. 5), we evaluated cell growth upon transfection of a miR-200a mimic. MiR-200a promoted cell growth, recapitulating the effect of KEAP1 silencing (Fig. 6A; Supporting Fig. 6A). The effect of miR-200a modulation was confirmed by its ability to down-regulate KEAP1 both in human and rat HCC cells (Fig. 6C; Supporting Fig. 6C) and to promote the expression of NRF2 target genes (Fig. 6B; Supporting Fig. 6B). Altogether, these results suggest that miR-200a controls the NRF2 pathway, whose activation promotes liver cancer cell growth.

Inhibition of NRF2 Is Associated With Regression of KRT-19-Positive Lesions

Since the role of NRF2 in early and/or intermediate stages of HCC development is still unknown, we aimed at investigating the effect of NRF2 modulation in these stages, characterized by a remarkable NRF2 activation (Fig. 3). Our previous studies have shown that a 7-day treatment with thyroid hormone (T3) is able to cause a significant reduction in the number of preneoplastic hepatic lesions in rats previously exposed to the R-H model.[22] Based on the above findings, we wished to determine whether the antitumorigenic activity of T3 could be mediated by NRF2. To this aim, 9 weeks after DENA initiation nodule-bearing rats were fed T3 for 4 and 7 days. While T3 treatment for 7 days caused a 50% reduction in the number of preneoplastic lesions compared to untreated animals, no difference between the two groups was observed at day 4 (Fig. 7A). We evaluated the expression of NRF2 and KEAP1 in microdissected KRT-19+ lesions from 4-day treated or untreated rats, namely, at a time that preceded the loss of preneoplastic nodules. T3 treatment resulted in global reduction of NRF2 expression and in loss of its nuclear localization (Fig. 7B), suggesting that T3 inactivates the NRF2-dependent pathway. This was confirmed by the down-regulation of NRF2 target genes (Fig. 7C). In parallel with NRF2 reduction, we observed an increase of KEAP1, which returned to the levels of the surrounding tissue. (Fig. 7B). In vitro experiments confirmed the inhibitory effect of T3 on the NRF2-dependent pathway (Supporting Fig. 7).

image

Figure 7. In vivo inhibition of the NRF2 pathway is followed by regression of KRT-19+ preneoplastic lesions. (A) Number of hepatic GSTP-positive preneoplastic lesions after 4 or 7 days of T3 treatment in rats previously exposed to the R-H model. The histogram represents mean values ± SE of 4 to 6 animals per group. NT: not treated. NS: not significant. ***P < 0.001. (B) IHC staining of NRF2 and KEAP1 in preneoplastic lesions of untreated (NT) or 4 days T3-treated animals. The insets show staining of nodules and normal surrounding tissue at lower magnification. Each panel is representative of five samples (magnification ×20; inset magnification ×4). (C) qRT-PCR analysis of NRF2 target gene expression in normal livers (control), KRT-19+ preneoplastic lesions either not treated (NT) or after 4 days of T3 treatment. Values are shown as fold change difference compared to normal liver. *P < 0.05.

Download figure to PowerPoint

Expression Signature of HCC Developed by the R-H Model Recapitulates That of Human HCC

To assess the translational value of our results, we determined how many differentially expressed genes and miRNAs in rat HCC are dysregulated also in human HCC. After removal of features not annotated on standard rat gene symbols from the gene array list, we extracted 159 genes from the original list of 234 dysregulated ones in rat HCCs. According to the Ensembl Database (V66; http://www.ncbi.nlm.nih.gov/pubmed/22086963), 110 genes out of 159 were rat-human orthologous (see Supporting Fig. 8 for experimental scheme). By integrating our data with the datasets annotated in Liverome, a curated database of liver cancer-related gene signatures,[23] we found that 78% (86/110) of the orthologous genes were dysregulated in HCC of both species (Supporting Table 5). Among these, 49 genes were dysregulated in a similar fashion in at least 50% of the works included in Liverome; moreover, 18 genes had the same type of dysregulation in at least one included work. Overall, the concordance between dysregulated genes in rat and human scored 78%, supporting the translational value of this model.

To determine which genes were already dysregulated at the beginning of the murine carcinogenic process, we intersected the 86 altered genes in HCCs in both species with those modified in KRT-19+ early nodules. Remarkably, we found that 65 genes out of 86 were dysregulated in rat preneoplastic lesions as well (bold in Supporting Table 5).

Literature-based analysis showed that 26 out of 46 miRNAs were commonly altered in rat and human HCC (Supporting Table 6) and that 10 out of these 26 miRNAs were already dysregulated in KRT-19-positive early lesions (bold in Supporting Table 6).

Altogether, these results clearly support the potential utility of the R-H model for detecting molecular alterations occurring in human HCC and identifying those arising at the onset of the process and likely critical for cancer development.

Discussion

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Discussion
  5. Acknowledgment
  6. References
  7. Supporting Information

Hepatocarcinogenesis is a slow process whereby the accumulation of genetic and epigenetic alterations eventually leads to the emergence and the expansion of clonal populations of transformed hepatocytes that evolve toward HCC. However, in humans the precise sequence of molecular events involved in tumor initiation and progression is not well defined, due to the limited access to early stages of tumor development. Thus, most of the studies have focused on fully developed HCCs and, as a consequence, information about the molecular alterations of early preneoplastic lesions is scanty. Since the molecular pathogenesis of HCC cannot be fully understood without more comprehensive knowledge of the molecular changes occurring during its early development, the histological precursors of human HCC—foci of phenotypically altered hepatocytes and dysplastic hepatocytes—deserve particular attention.[24] In the current work we analyzed stepwise global miRNA and mRNA expression in a rat model of HCC, with emphasis on the very early steps characterized by the occurrence of phenotypically altered preneoplastic nodules.

One of the most relevant findings stemming from our work is that a number of miRNAs are already dysregulated in KRT-19+ preneoplastic nodules. Since these lesions are considered the HCC precursors in the carcinogenesis model used in the present study,[11] it is likely that these miRNAs play a relevant role in HCC onset. The identification of miRNAs altered at the beginning of the carcinogenic process is a novel finding, since very few contributions have attempted to address the impact of miRNA dysregulation at this stage of HCC development. Indeed, previous studies aimed at identifying miRNA alterations at the beginning of hepatocarcinogenesis have evaluated miRNA expression only in the whole liver of mice exposed to a carcinogenic regimen—characterized by hepatic fat accumulation and inflammatory response (the choline-devoid methionine deficient model)—before the appearance of preneoplastic lesions, rather than in isolated nodules.[25, 26]

Among the miRNAs found dysregulated in our study, some have been reported as modified in human HCC, while others have not been previously associated with liver cancer. Although further studies are warranted to better define the role of these miRNAs and of their targets, they might represent novel critical players in the development and progression of HCC. In particular, the present study identified 13 miRNAs that are dysregulated from the very early stages of the carcinogenic process throughout the progression to HCC, suggesting that they participate in the initial events leading to HCC development and that are required for neoplastic progression. Among these miRNAs, miR-224, miR-125b, miR-375, and miR-122 had already been identified as dysregulated in human HCC,[7, 9, 27, 28] whereas others, such as miR-802, miR-429, and miR-499 have not been previously described.

A second important finding is that 85% of the most up-regulated and 80% of the most down-regulated genes in rat HCC were already altered in early KRT-19+ preneoplastic nodules. Remarkably, an impressive number of genes involved in xenobiotic metabolism and NRF2-mediated oxidative stress signaling pathway were modified from the beginning of the tumorigenic progress. This is very relevant, as it suggests that metabolic changes are likely necessary, although not sufficient, to allow the upsurge of preneoplastic lesions and to sustain the progression of early lesions to a malignant condition. This metabolic readjustment might be the consequence of a coordinated survival response to the DENA/2-acetylaminofluorene (2-AAF) induced-damage. However, since a similar metabolic shift has been observed also in a transgenic mouse model in the absence of chemical treatment[29] and in human HCC,[30] a response to oxidative stress appears to be a general feature caused by hostile conditions, such as viruses, alcohol, or chemicals. Overall, these data suggest that the antioxidant response is an initial adaptive mechanism triggered to assure a better survival for liver cells, becoming irreversibly altered in some of them (initiated cells?). The finding that in the R-H model this metabolic derangement persists after the first 4 weeks of treatment, when the chemical-induced stress is no longer present, and the observation that dysregulation of the same genes is observed in advanced HCC indicate that these initial changes may likely play a far more critical role than previously recognized.

Another interesting observation is that the comparison between dysregulated genes and miRNAs in each step of progression showed the existence of networks where a consistent percentage of modified genes are indeed targeted by dysregulated miRNAs. In the transition from normal liver to KRT-19+ nodules we identified two relevant nodes where miR-30 and miR-200 families control the expression of many modified genes. While miR-30a, -30d, and -30e were modified mainly at this initial stage, miR-200a, -200b, and -429 were up-regulated at both the initial and late stages. It is worth noting that miR-200a is known to negatively regulate the NRF2 pathway, which is already activated in early preneoplastic KRT-19+ lesions. Following exposure of cells to electrophiles or oxidative stress, NRF2 is able to escape KEAP1-mediated degradation, translocate to the nucleus, and activate the expression of a series of antioxidative and cytoprotective proteins including NQO1, GCLC, and several members of the GST family.[31] Notably, the role of this transcription factor has recently become the topic of an important controversy, since it is unclear whether NRF2 acts as a tumor suppressor or as an oncogene.[17] Indeed, while many studies suggest that NRF2 activation mediates the beneficial effects of chemopreventive drugs, genetic analyses of human tumors indicate that this transcription factor may exert an oncogenic effect and cause resistance to chemotherapy.[32] Moreover, a recent work has identified putative activating mutations of this gene in 6.4% of HCCs, further sustaining its oncogenic role.[33] However, as recently reviewed,[17] the NRF2 role in early/intermediate steps of the tumorigenic process is largely unknown. Our study provides evidence of an oncogenic role of NRF2 in preneoplastic/premalignant stages of hepatocarcinogenesis. Indeed: (1) many NRF2 target genes were among the most up-regulated; (2) many members of the small MAF family of NRF2 coactivators were found activated; (3) NRF2 silencing impaired liver cancer cell proliferation in vitro; and (4) in vivo treatment of rats with T3 caused the inhibition of the NRF2 pathway followed by the regression of KRT-19+ preneoplastic lesion.

Finally, the present study shows the common dysregulation of many miRNAs and genes in both rat and human HCC. Such a striking similarity in miRNA/gene expression perturbation among HCC from different species and etiologies demonstrates that the R-H model recapitulates human HCC to an extent that allows successful mining of target genes. In this line, our results indicating an oncogenic role of the NRF2 pathway in preneoplastic lesions should be considered for their possible translational value. Further investigation is warranted to verify NRF2 activation in human preneoplastic stages as well. If so, targeting this pathway would offer new therapeutic options in stages of progression that could dramatically change the evolution of the disease.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Discussion
  5. Acknowledgment
  6. References
  7. Supporting Information

We thank G. Diaz for support in statistical analysis, M. Angioni and A. Follenzi for generation and characterization of RH cells, B. Martinoglio for qRT-PCR, and F. Natale for editing the article.

References

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Discussion
  5. Acknowledgment
  6. References
  7. Supporting Information
  • 1
    Bosch FX, Ribes J, Cleries R, Diaz M. Epidemiology of hepatocellular carcinoma. Clin Liver Dis 2005;9:191-211, v.
  • 2
    Hoshida Y, Toffanin S, Lachenmayer A, Villanueva A, Minguez B, Llovet JM. Molecular classification and novel targets in hepatocellular carcinoma: recent advancements. Semin Liver Dis 2010;30:35-51.
  • 3
    Lee JS, Thorgeirsson SS. Genome-scale profiling of gene expression in hepatocellular carcinoma: classification, survival prediction, and identification of therapeutic targets. Gastroenterology 2004;127:S51-S55.
  • 4
    Calin GA, Croce CM. MicroRNA signatures in human cancers. Nat Rev Cancer 2006;6:857-866.
  • 5
    He L, Thomson JM, Hemann MT, Hernando-Monge E, Mu D, Goodson S, et al. A microRNA polycistron as a potential human oncogene. Nature 2005;435:828-833.
  • 6
    Johnson SM, Grosshans H, Shingara J, Byrom M, Jarvis R, Cheng A, et al. RAS is regulated by the let-7 microRNA family. Cell 2005;120:635-647.
  • 7
    Giordano S, Columbano A. MicroRNAs: New tools for diagnosis, prognosis, and therapy in hepatocellular carcinoma? Hepatology 2013;57:840-847.
  • 8
    Imbeaud S, Ladeiro Y, Zucman-Rossi J. Identification of novel oncogenes and tumor suppressors in hepatocellular carcinoma. Semin Liver Dis 2010;30:75-86.
  • 9
    Jiang J, Gusev Y, Aderca I, Mettler TA, Nagorney DM, Brackett DJ, et al. Association of MicroRNA expression in hepatocellular carcinomas with hepatitis infection, cirrhosis, and patient survival. Clin Cancer Res 2008;14:419-427.
  • 10
    Solt DB, Medline A, Farber E. Rapid emergence of carcinogen-induced hyperplastic lesions in a new model for the sequential analysis of liver carcinogenesis. Am J Pathol 1977;88:595-618.
  • 11
    Andersen JB, Loi R, Perra A, Factor VM, Ledda-Columbano GM, Columbano A, et al. Progenitor-derived hepatocellular carcinoma model in the rat. Hepatology 2010;51:1401-1409.
  • 12
    Lee JS, Heo J, Libbrecht L, Chu IS, Kaposi-Novak P, Calvisi DF, et al. A novel prognostic subtype of human hepatocellular carcinoma derived from hepatic progenitor cells. Nat Med 2006;12:410-416.
  • 13
    Petrelli A, Perra A, Schernhuber K, Cargnelutti M, Salvi A, Migliore C, et al. Sequential analysis of multistage hepatocarcinogenesis reveals that miR-100 and PLK1 dysregulation is an early event maintained along tumor progression. Oncogene 2012;31:4517-4526.
  • 14
    Smyth GK. Limma: linear models for microarray data. In: R.Gentleman VCSDRIWH, ed. Bioinformatics and Computational Biology Solutions using R and Bioconductor. New York: Springer, 2005. p 397-420.
  • 15
    Borel F, Konstantinova P, Jansen PL. Diagnostic and therapeutic potential of miRNA signatures in patients with hepatocellular carcinoma. J Hepatol 2012;56:1371-1383.
  • 16
    Malz M, Pinna F, Schirmacher P, Breuhahn K. Transcriptional regulators in hepatocarcinogenesis—key integrators of malignant transformation. J Hepatol 2012;57:186-195.
  • 17
    Sporn MB, Liby KT. NRF2 and cancer: the good, the bad and the importance of context. Nat Rev Cancer 2012;12:564-571.
  • 18
    Marini MG, Chan K, Casula L, Kan YW, Cao A, Moi P. hMAF, a small human transcription factor that heterodimerizes specifically with Nrf1 and Nrf2. J Biol Chem 1997;272:16490-16497.
  • 19
    Kobayashi A, Kang MI, Okawa H, Ohtsuji M, Zenke Y, Chiba T, et al. Oxidative stress sensor Keap1 functions as an adaptor for Cul3-based E3 ligase to regulate proteasomal degradation of Nrf2. Mol Cell Biol 2004;24:7130-7139.
  • 20
    Yeager RL, Reisman SA, Aleksunes LM, Klaassen CD. Introducing the “TCDD-inducible AhR-Nrf2 gene battery.” Toxicol Sci 2009;111:238-246.
  • 21
    Eades G, Yang M, Yao Y, Zhang Y, Zhou Q. miR-200a regulates Nrf2 activation by targeting Keap1 mRNA in breast cancer cells. J Biol Chem 2011;286:40725-40733.
  • 22
    Ledda-Columbano GM, Perra A, Loi R, Shinozuka H, Columbano A. Cell proliferation induced by triiodothyronine in rat liver is associated with nodule regression and reduction of hepatocellular carcinomas. Cancer Res 2000;60:603-609.
  • 23
    Lee L, Wang K, Li G, Xie Z, Wang Y, Xu J, et al. Liverome: a curated database of liver cancer-related gene signatures with self-contained context information. BMC Genomics 2011;12(Suppl 3):S3.
  • 24
    Thorgeirsson SS, Grisham JW. Molecular pathogenesis of human hepatocellular carcinoma. Nat Genet 2002;31:339-346.
  • 25
    Wang B, Majumder S, Nuovo G, Kutay H, Volinia S, Patel T, et al. Role of microRNA-155 at early stages of hepatocarcinogenesis induced by choline-deficient and amino acid-defined diet in C57BL/6 mice. Hepatology 2009;50:1152-1161.
  • 26
    Wang B, Hsu SH, Majumder S, Kutay H, Huang W, Jacob ST, et al. TGFbeta-mediated upregulation of hepatic miR-181b promotes hepatocarcinogenesis by targeting TIMP3. Oncogene 2010;29:1787-1797.
  • 27
    Ladeiro Y, Couchy G, Balabaud C, Bioulac-Sage P, Pelletier L, Rebouissou S, et al. MicroRNA profiling in hepatocellular tumors is associated with clinical features and oncogene/tumor suppressor gene mutations. Hepatology 2008;47:1955-1963.
  • 28
    Budhu A, Jia HL, Forgues M, Liu CG, Goldstein D, Lam A, et al. Identification of metastasis-related microRNAs in hepatocellular carcinoma. Hepatology 2008;47:897-907.
  • 29
    Coulouarn C, Factor VM, Conner EA, Thorgeirsson SS. Genomic modeling of tumor onset and progression in a mouse model of aggressive human liver cancer. Carcinogenesis 2011;32:1434-1440.
  • 30
    Inami Y, Waguri S, Sakamoto A, Kouno T, Nakada K, Hino O, et al. Persistent activation of Nrf2 through p62 in hepatocellular carcinoma cells. J Cell Biol 2011;193:275-284.
  • 31
    Baird L, nkova-Kostova AT. The cytoprotective role of the Keap1-Nrf2 pathway. Arch Toxicol 2011;85:241-272.
  • 32
    Wang XJ, Sun Z, Villeneuve NF, Zhang S, Zhao F, Li Y, et al. Nrf2 enhances resistance of cancer cells to chemotherapeutic drugs, the dark side of Nrf2. Carcinogenesis 2008;29:1235-1243.
  • 33
    Guichard C, Amaddeo G, Imbeaud S, Ladeiro Y, Pelletier L, Maad IB, et al. Integrated analysis of somatic mutations and focal copy-number changes identifies key genes and pathways in hepatocellular carcinoma. Nat Genet 2012;44:694-698.

Supporting Information

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Discussion
  5. Acknowledgment
  6. References
  7. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
hep26616-sup-0001-suppfig1.tif4140KFigure S1. MiRNA profile classifies the different steps of hepatocarcinogenesis. Hierarchical clustering of miRNA expression profile in KRT-19- and KRT-19+ preneoplastic lesions, adenomas (Ade), early HCCs (eHCCs) and advanced HCCs (aHCCs). Only miRNAs whose expression was dysregulated more than 4-fold in a minimum of two samples in at least one experimental group were considered. Each row represents the expression profile of a miRNA expressed as ΔΔ ct calculated respect to age-matched controls (see Supporting Material for details) and each column represents a sample. Red and green colors represent higher or lower expression levels of the miRNA (median centered), respectively.
hep26616-sup-0002-suppfig2.tif200KFigure S2. QRT-PCR validation of randomly selected miRNAs in the stepwise development of HCC. MiRNA expression was determined in KRT-19- and KRT-19+ nodules, adenomas, eHCCs and aHCCs. MiRNA expression is reported as log fold-change relative to age-matched controls. TLDA: TaqMan Low Density Array.
hep26616-sup-0003-suppfig3.tif178KFigure S3. QRT-PCR validation of randomly selected genes in the stepwise development of HCC. Dynamic mRNA expression was determined in KRT-19- and KRT-19+ nodules, adenomas, eHCCs and aHCCs. Gene expression is reported as log fold-change relative to age-matched controls.
hep26616-sup-0004-suppfig4.tif3344KFigure S4. Network visualization of the indicated transcription factors. Data and visualization were obtained using IPA. Red: up-regulated genes; green: downregulated genes. TF → target gene.
hep26616-sup-0005-suppfig5.tif6612KFigure S5. The miR-200 and miR-30 families are dysregulated during hepatocarcinogenesis. MiR-200 (A) and miR-30 (B) family expression evaluated by TLDA in different stages. MiRNA expression is reported as log fold-change compared to age-matched controls. TLDA: TaqMan Low Density Array.
hep26616-sup-0006-suppfig6.tif6612KFigure S6. Modulation of the NRF2 pathway affects HCC cell growth. A) HepG2 cells were transiently transfected with NRF2 or KEAP1 or control (siC) siRNAs or a miR-200a mimic and seeded in 96-well plates. Cell growth was evaluated by crystal violet staining 72 hours after seeding, in either the presence (H2O2) or the absence (NT) of 25uM hydrogen peroxide. B) Expression of the NRF2 target genes NQO1, GCLC and GSTA4 was evaluated by qRT-PCR on RNA extracted from the cells transfected in (A). Values are shown as fold change difference compared to control transfected cells (siC). C) HepG2 and rat cells (FaO and RH) were transiently transfected with the indicated siRNAs or miR-200a. Total cell lysates were extracted 48 hours upon transfection and analyzed by Western blot. Vinculin or βactin were used as loading control. Quantification of KEAP1 related to the corresponding βactin is shown for miR-200a (control siC was set as 1).
hep26616-sup-0007-suppfig7.tif72KFigure S7. In vitro T3 treatment of HepG2 cells inhibits the NRF2 pathway. HepG2 cells were treated with 10-6 M triiodothyronine for 48 hours. Total RNA was then extracted and expression of NRF2 target genes (NQO1, GCLC, GSTA4) was evaluated. Data are indicated as fold changes compared to not treated cells (NT).
hep26616-sup-0008-suppfig8.tif712KFigure S8. Flowchart for the intersection with the Liverome database. We first downloaded gene signatures from the Liverome database, for a total of 6.927 human genes. Then we focused on the dataset of 234 rat genes dysregulated in our model (aHCC vs ctr). For these genes, we looked for reliable human orthologous by means of the Ensembl database (version 66). Commonly dysregulated genes were then identified according to matching gene ids. Up/Down common modulations were assessed by means of the Fold-Change provided by the Limma package for the rat case and by the internal Liverome annotation for the human case.
hep26616-sup-0009-supptable1.doc29KSupp. Table 1. Percentage of early preneoplastic nodules (10 weeks after initiation with DENA) and aHCCs (14 months after initiation) positive for KRT-19.
hep26616-sup-0010-supptable2.doc78KSupp. Table 2A and Table 2B.
hep26616-sup-0011-supptable3.doc67KSupp. Table3A, Table3B, Table3C and Table3D.
hep26616-sup-0012-supptable4.doc36KSupp. Table 4. MiRNAs differentially expressed in KRT-19+ vs. KRT-19-
hep26616-sup-0013-supptable5.doc112KSupp. Table 5. List of the 86 rat/human orthologous genes involved in HCC progression
hep26616-sup-0014-supptable6.doc52KSupp. Table 6. List of the 26 miRNAs involved in both rat and human HCC progression
hep26616-sup-0015-suppinfo.doc67KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.