JNK1 activation predicts the prognostic outcome of the human hepatocellular carcinoma
© Chang et al; licensee BioMed Central Ltd. 2009
Received: 19 April 2009
Accepted: 17 August 2009
Published: 17 August 2009
Skip to main content
© Chang et al; licensee BioMed Central Ltd. 2009
Received: 19 April 2009
Accepted: 17 August 2009
Published: 17 August 2009
Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide with an extremely poor prognosis. The classification of HCC based on the molecular signature is not well-established.
In the present study, we reported HCC signature genes based on the JNK1 activation status in 31 HCC specimens relative to the matched distal noncancerous liver tissue from 31 patients. The HCCs with high JNK1 (H-JNK1) and low JNK1 (L-JNK1) were sub-grouped. Two different signature gene sets for both H-JNK1 and L-JNK1 HCC were identified through gene expression profiling. A striking overlap of signature genes was observed between the H-JNK1 HCC and the hepatoblastoma or hepatoblastoma-type HCC. Many established biomarkers for hepatic progenitor cells were over-expressed in H-JNK1 HCC, including AFP, TACSTD1, KRT19, KRT7, THY1, and PROM1. In addition, the majority of the most up-regulated genes were those associated with metastasis and earlier recurrence, whereas the genes for normal liver function were substantially down-regulated in H-JNK1 HCC tissue. A Kaplan-Meier plot demonstrated that the survival of the patients with H-JNK1 HCC was severely impaired.
Accordingly, we believe that the H-JNK1 HCC may originate from hepatic progenitor cells and is associated with poorer prognosis. The status of JNK1 activation in HCC tissue, thus, might be a new biomarker for HCC prognosis and therapeutic targeting.
Hepatocellular carcinoma (HCC) has plagued populations in Far East Asia, South Asia and Sub-Saharan Africa for several decades where the prevalence of hepatitis B viral (HBV) infection and aflatoxin exposure is high[1, 2]. A sharp increase in HCC incidence in North America, Western Europe and Japan has been noted in recent years due to hepatitis C viral (HCV) infection, alcohol abuse and non-alcoholic fatty liver disease [3–6]. HCC is the fourth most common neoplasm and the third most common cause of cancer-related death worldwide. Age-adjusted HCC incidence rates vary from 2 per 100,000 population in North America to 80 per 100,000 population in China. Since most HCC patients are diagnosed when the tumors are in an advanced stage and the majority of HCCs develop in the context of chronic liver cirrhosis, curative therapy is not available yet[7, 8]. The overall 5-year survival rate is lower than 5% . Accordingly, a new understanding of the molecular mechanisms controlling the development and progression of HCC is urgently needed for the efficient treatment of this deadly disease.
A major goal in current HCC research is to define gene signatures that drive initiation, maintenance and progression of the malignant tumors, which is anticipated to be helpful in classifying tumor stages and predicting prognostic outcomes, such as metastasis, patient survival rate and recurrence of the tumors after resection. It is also highly desirable to use the gene signature to design targeted therapies or the so-called personalized medicine . However, amid this euphoria, the HCC signature genes identified thus far vary considerably, depending on a number of etiological and pathological features, including viral infection, inflammation, cirrhosis, necrosis, fibrosis, vascular invasion, tumor cell origination, and environmental exposure. The situation is further complicated by the byzantine complexity of algorithms for data analysis and the use of different platforms for gene profiling [9–12]. These factors might explain why so many reports concerning HCC signature genes have resulted in therapeutic dead ends.
Accumulating evidence suggests that activation of the protein kinases provides growth advantage to cells during malignant transformation and tumorigenesis. Recently, we[13, 14] and Hui et al. simultaneously demonstrated critical role of JNK1 activation in the pathogenesis of human HCC and/or a mouse model of chemical carcinogen-induced HCC. Both studies observed overactivation of JNK1 in about 55% and 56% human HCC relative to the noncancerous liver tissue, respectively[13, 15]. We have shown an association between JNK1 activation and overexpression of the genes for cell growth, histone methylation, and the downregulation of the genes for cell differentiation, antioxidant defense and drug or lipid metabolism. By re-analyzing and comparing our gene profiling data with the signature genes for HCC with a poor prognosis and hepatoblastoma (HB) reported recently[9, 16], we found a striking overlap of the gene expression pattern in high JNK1 (H-JNK1) HCC tissue with the signature genes for HCC with a poor prognosis and HB, respectively. Assessment of the patient survival data indicates that the overall survival of the patients with H-JNK1 HCC is substantially impaired. JNK1, rather than JNK2, has been implicated in development of steatohepatitis[17, 18] and carcinogen-induced mouse HCC. It is very likely, thus, that JNK1 is a critical contributor to the progression of human HCC, which can serve as a potential target for new therapies.
Clinical information for the H-JNK1 and L-JNK1 HCC patients
n = 17
n = 14
p – value
Age, y (range)
Tumor grade (II/III)
Tumor diameter (cm)
7.3 ± 2.8
5.0 ± 2.7
21.7 ± 19**
39.1 ± 30
14 ± 14
42 ± 30
43 ± 14
28 ± 34
The 10 most up- and down-regulated genes in HCC tissue vs ANC tissue
ribonucleotide reductase M2
aurora kinase A
nucleolar/spindle protein 1
family with similarity 83 D
MLF1 interacting protein
epithelial transforming 2
carnosine dipeptidase 1
complement component 7
complement component 9
aldo-keto reductase 1D1
oncoprotein induced transcript 3
solute carrier organic transporter
hepcidin antimicrobial peptide
complement factor H-related 3
complement factor H-related 4
An additional functional comparison between these two groups indicates that the expression of the genes for amino acid metabolism, small molecule biochemistry, cell cycle, and DNA replication was overexpressed in the H-JNK1 HCC tissue relative to the L-JNK1 HCC tissue (Fig. 4B). In contrast, the genes for cell-to-cell interaction, cell signaling, cell morphology, and free radical scavenging were up-regulated in the L-JNK1 HCC tissue relative to the H-JNK1 HCC tissue (Fig. 4B).
Overlapping of H-JNK1 signature genes with other types of HCCs
Up In HB**
Fifteen out of the 30 most down-regulated genes in the H-JNK1 HCC tissue were covered by those down-regulated genes in the subclass A HCC reported by Lee et al. The subclass A HCC had been later confirmed as the HCCs derived from the hepatic progenitor cells with poorer survival. These genes include HPD, CYP3A4, SERPINC1, AQP9, GLYAT, OTC, PCK1, CYP2E1, CYP2C9, CYP8B1, FMO3, BAAT, SLC27A5, SLC22A1, and CYP2D6 (Table 3). In addition, 9 genes, including GYS2 , F9 [12, 35], NAT2 [11, 36, 37], ADH1B , LECT2 [11, 38, 39], AFM , ADH4 [11, 21], SLC10A1 , and HSD11B1 , in this down-regulated gene list that were not covered in the down-regulated genes of Lee et al., had been recently or previously reported to be substantially decreased in HCC with poorer prognosis.
The similarity between HCCs with higher JNK1 activation and the HCCs with characteristics of hepatic progenitor cells (Fig. 3A) suggests to us that the H-JNK1 HCCs possibly have features of HB. The signature genes for HB has been recently reported by Cairo et al . By comparing the signature genes of HB with that of H-JNK1 HCC tissue, a striking similarity between the genes in robust Cluster 2 (rC2) HB and the genes associated with the JNK1 activation status was noted. For the top 400 increased and greatest 100 decreased genes in rC2 HB (Supplemantary table S7 of Cairo et al ), 90% and 94% of these genes are presented among the most increased and decreased expressed genes in the H-JNK1 HCC tissue, respectively. Vice versa, 11 out of the 30 most up-regulated genes and 20 out of the 30 most down-regulated genes in the H-JNK1 HCC tissue are represented by the up- and down-regulated genes in the HB or rC2 HB, respectively (Table 3). A significant agreement of the 16 signature genes differentiating rC2 and rC1 HB (see Cairo et al ) and the signature genes for the H-JNK1 HCC tissue was evident. The listed up- and down-regulated 16 signature genes in rC2 HB are fully overlap with the up- and down-regulated genes in H-JNK1 HCCs (Additional file 9).
We also noted that similar to the rC2 HB, several liver progenitor markers, such as AFP , TACSTD1 , KRT19 [16, 22], KRT7 , PROM1 (CD133) [41, 42], THY1 (CD90)  and VIM , are significantly up-regulated in the H-JNK1 HCC samples relative to the L-JNK1 HCC samples (Additional file 10). A characteristic feature of the HB and hepatic progenitor cells is the enhanced expression of the imprinted genes, including H19, IGF2, DLK1, PEG3, PEG10, MEG3, SGCE, and NDN . Indeed, all of these mentioned imprinted genes are up-regulated in the H-JNK1 HCC samples (Additional file 10).
HCC is a heterogeneous type of tumor despite overall outcomes of poor prognosis. Depending on etiological and the accompanying pathological conditions, the survival or tumor recurrence vary considerably among HCC patients. The cellular origin of HCC may play key roles in determining the status of metastasis, intra-hepatic recurrence and patient survival. It has been debated for decades whether HCC arises exclusively by dedifferentiation of mature hepatocytes or maturation arrest of liver stem cells or progenitor cells (oval cells in rodents). Great efforts have been made in identifying bipotential hepatic progenitor cells that can differentiate into hepatocytes and cholangiocytes . Markers for hepatocytes, such as AFP and albumin, and cholangiocytes, mainly KRT7 and KRT19, are expressed in the progenitor cells. Evidence supporting progenitor cell origin of HCC is largely based on the fact that many HCCs concurrently express both hepatocytic markers and cholangiocytic markers along with stem cell markers including TACSTD1 (EpCAM), PROM1 (CD133) and THY1 (CD90) [41, 42]. The observation that a profound concordance of gene profiles for H-JNK1 HCC tissue with those of Lee et al. on HB-type HCC tissue and Cairo et al.  on HB samples may support the hypothesis that H-JNK1 HCC is derived from the liver progenitor cells. Indeed, the genes for KRT7, KRT19, PROM1, AFP, as well as imprint genes are overexpressed in the H-JNK1 HCC tissue (Additional file 10).
The HB-like features of the H-JNK1 HCC tissue are additionally supported by the recent characterization of the EpCAM positive cells in human HCC samples. By classifying 235 HCC tumor specimens with gene profiling and immunohistochemistry analyses, Yamashita et al. reported signature genes that were differentially expressed between EpCAM positive, stem cell-like HCC (HpSC-HCC) and the EpCAM negative mature hepatocyte-like HCC. A number of genes are commonly over-expressed between H-JNK1 HCC and the HpSC-HCC, such as S100P, S100A9, CCL20, PEG10, CD24, NTS, H19, UBE2C, TMED3, and KRT19 that were listed among the 30 most up-regulated genes in H-JNK1 HCC tissue (Table 3). Majority of the 30 most down-regulated genes, 22 genes, were also under-expressed in HpSC-HCC tissue. Interestingly, both H-JNK1 HCC and HpSC-HCC samples showed poorer prognosis in a Kaplan-Meier survival analysis.
Most recently, Villanueva et al. reported signature genes for HCC samples with an aberrant mTOR signaling. About 40% of the 193 up-regulated genes in this type of HCC were found in the up-regulated gene list for H-JNK1 HCC samples. The coverage was increased to 86% if one relaxed the 5-fold cutoff to 1.5 fold and removed those uncharacterized genes (data not shown). For the down-regulated genes, more than 91% of the known genes were represented in the list of the down-regulated genes in H-JNK1 HCC samples. Vice versa, the genes for CCL20, CD24, GALNT7, AMIGO2, AFP, UBE2C, and SOX4 in the 30 most up-regulated genes in the H-JNK1 HCC samples were represented in the HCC samples with an aberrant activation of mTOR signaling (Table 3). The similarity in gene expression pattern between H-JNK1 HCC and the HCC with abnormal mTOR signaling further supports the assumption that H-JNK1 HCCs are the tumors with a highly aggressive nature and poorer prognosis.
It is worth to mention that there appears to be a gender disparity in JNK1 activation and HCC survival probability between male and female patients (Fig. 6B). A strong correlation between higher JNK1 activation and lower survival probability was noted in male, but not in female patients. Such disparity may be possibly attributed to the different levels of the sex steroids between two genders. It has long been hypothesized that certain female sex hormone, esp. estrogen, is repressive for the expression of inflammatory genes, such as IL-6, leading to tumor suppression . In contrast, male sex hormone, androgen, appears to be a promoting factor for HCC . It is unclear at the present whether the higher JNK1 activity or its down-stream effect in the female HCC patients is antagonized by the sex hormone. Nevertheless, the observed disparity in JNK1 activation and HCC prognosis between male and female patients is compensatory to the human epidemiological and animal tumorigenic studies demonstrating that the HCC incidence rate in male is about 3–5 times higher than that in female [8, 45].
In summary, we believe that the present report not only provides evidence indicating that JNK1 activation contributes to poorer HCC prognosis but also aids in molecular classification for HCC. It is conceivable that HCC samples with high JNK1 activation possess feature of hepatic progenitor cells and have similarities with HB, HB-type HCC, mTOR-altered HCC, and the EpCAM+/AFP+ HCC (HpSC-HCC). In addition to this notion, complementary evidence to support JNK1 activation in a subset of HCC tissue was previously provided by Lee et al. who demonstrated a strong enrichment of the AP-1 signaling, a direct down-stream target of JNK1, in HB-type HCC samples. The final goal of cancer gene profiling is to design targeted therapy and/or so-called personalized medicine. However, such goal is hard to achieve currently due to the large quantity of the cancer signature genes and the lack of a clear picture of the drug targetable genes or gene networks. The activation of the major signaling pathways, such as JNK1 and mTOR, on the other hand, is potentially targetable by relatively specific therapeutic agents. Thus, the identification of H-JNK1 HCC opens up possibilities to develop new treatment strategies targeting the intercellular signaling to delay the progression of HCC.
The human HCC tissue samples were collected from patients undergoing surgical resection procedures with informed consent from the patients according to a protocol approved by the Institutional Review Board at Qidong Liver Cancer Institute, Qidong, PR China. Partial adjacent noncancerous (ANC) liver tissue with an average distance of 1.28 cm from the tumor were also collected for pathological evaluation. The tissue specimens were stored in liquid nitrogen immediately for later pathological and biological analyses. The diagnostic, clinical and pathological stages of the tumors were determined as previously reported. Clinical relevant information for the HCC patients is shown in Table 1. A total of 40 HCC samples along with 40 case-matched ANC liver tissues were collected, among which 31 HCC samples and the matched ANC tissue were used in the present study. The other 9 HCC samples were excluded from this study because of combined cholangiocarcinoma or other type of tumors. For immunohistochemistry (IHC) studies, tissue microarray slides containing total of 52 HCC samples were purchased from US Biomax, Inc., (Rockville, MD), (See following).
The JNK activation in both HCC tissues and non-cancerous liver tissues was determined through Western blotting using antibodies against phospho-JNK and total JNK purchased from Cell Signaling (Beverly, MA) as described previously. JNK1 specific kinase activity assay was performed through immunoprecipitation using JNK1 antibody purchased from Santa Cruz Biotechnology Inc. (sc-474, Santa Cruz, CA) and the JNK kinase activity assay kit provided by Cell Signaling Technology (Beverly, MA). Tissue microarray slides (LV241 and LV801) purchased from US Biomax, Inc. (Rockville, MD) were used for IHC. LV241 and LV801 slides contained 12 and 40 HCC specimens along with the case-matched ANC tissue, respectively. The information about the histopathological and patient data is available at the manufacturer's website http://www.biomax.us/tissue-arrays/. The tissue slides were baked at 60°C for 30 min and then deparaffinized by xylene before IHC. For IHC, the slides were first subjected to antigen retrieval and then incubated with normal goat serum for 20 min at room temperature, followed by incubation with anti-phospho-JNK antibody (1:50 dilution). Other procedures of IHC were applied as suggested by the manufacturer. Specific antibodies for JNK1, JNK2 and phosphorylated-JNK were purchased from Santa Cruz Biotechnology (Santa Cruz, CA) or Cell Signaling Technology (Beverly, MA).
Gene expression profiling was performed according to the protocols described by Affymetrix, Inc. (Santa Clara, CA). Raw data from Affymetrix HG-U133 plus 2.0 Gene Chip microarray were normalized using the GeneChip Robust Multiarray Averaging (GCRMA) method and by removing probe sets with "absence" call on the array. The data were further calibrated by False Discovery Rate (FDR) correction with a p-value ≤ 0.05. The gene profiling data were further sub-grouped according to the JNK1 activation status in the HCC tissue relative to the ANC tissue to generate gene lists for both the high JNK1 (H-JNK1) HCC tissue and low JNK1 (L-JNK1) HCC tissue. A list of genes with a larger than 5-fold differential expression between H-JNK1 HCC and L-JNK1 HCC was also made based on direct comparison between these two groups. An online trial version of Ingenuity Pathways Analysis was used for the interactive network assay. For each gene list, independent analyses for all genes, over-expressed genes, and under-expressed genes were performed. The significance of each network was estimated as described by Lee et al..
Primers were designed using the universal probe library from Roche Applied Science (Applied Biosystems). Primer sequences and probe numbers are available upon request. Gene expression levels were analyzed by quantitative PCR in an ABI-PRISM® 7900 sequence detection system (Applied Biosystems) and normalized to GAPDH. Samples were run in triplicate under quantitative PCR conditions. Data were quantified and normalized using the ΔCt method. Cycling conditions PCR were 2 min at 50°C, 10 min at 95°C, then 40 cycles of 15 s at 95°C, and 1 min at 60°C.
Patient survival data were collected through the Qidong Cancer Registration System, a comprehensive cancer surveillance program, and follow-up monitoring of the patients after tumor resection. Patient survival time in months was calculated from the date of the first diagnosis of the tumor to the date of patient death. For the patients who were alive at the time of final data collection on September 30, 2008, the survival time was determined from the date of the first diagnosis to September 30, 2008. The Kaplan-Meier method was used to determine the survival probability of the patients with H-JNK1 HCC or L-JNK1 HCC. SigmaPlot 9.0 statistics software or Fisher's Exact Test was used to analyze statistic significance or clinical features.
Raw data of microarray can be found from: http://sites.google.com/site/cancerepigeneticsandsignaling/
We thank Dr. Xin Wei Wang, at the Liver Carcinogenesis Section, Laboratory of Human Carcinogenesis, NCI/NIH (Bethesda, MD), for sharing his unpublished cDNA array data with us for comparison. We also thank the members in the Biostatistics and Epidemiology Branch of NIOSH for the re-normalization of the gene profiling data. This research project was partially supported through NIOSH/CDC intramural funding (9270036) to F.C.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.