Identification of intrahepatic cholangiocarcinoma ...

BBRC Biochemical and Biophysical Research Communications 345 (2006) 1022–1032 www.elsevier.com/locate/ybbrc

Identification of intrahepatic cholangiocarcinoma related genes by comparison with normal liver tissues using expressed sequence tags Ai-Guo Wang a, Sun Young Yoon a, Jung-Hwa Oh a, Yeo-Jin Jeon a, Mirang Kim a, Jeong-Min Kim a, Sang-Soon Byun a, Jin Ok Yang b, Joo Heon Kim c, Dae-Ghon Kim d, Young-il Yeom a, Hyang-Sook Yoo a, Yong Sung Kim a, Nam-Soon Kim a,* a

b

Laboratory of Human Genomics, Genome Research Center, Korea Research Institute of Bioscience and Biotechnology (KRIBB), Daejeon, Republic of Korea National Genome Information Center, Korea Research Institute of Bioscience and Biotechnology (KRIBB), Daejeon, Republic of Korea c Department of Pathology, Eulji University School of Medicine, Daejeon, Republic of Korea d Department of Internal Medicine, Chonbuk National University Medical School and Hospital, Chonju, Republic of Korea Received 25 April 2006 Available online 11 May 2006

Abstract Intrahepatic cholangiocarcinoma (ICC), a malignant tumor derived from the bile duct epithelium, is one of the leading causes of death from cancer, worldwide. However, the mechanisms related to it remain largely unknown. In this study, an analysis of the gene expression profiles for ICC was done using the frequency of the ESTs obtained from nine cDNA libraries that constructed from 4 ICC cell lines and 4 normal liver tissues. One hundred and thirty-seven genes were identified as being either up- or down-regulated in human ICC cells. Thirty genes were randomly selected to confirm their differential expression in 4 human ICC cell lines and 5 ICC tissues compared to normal liver tissues by semi-quantitative RT-PCR. Among these genes, ANXA1, ANXA2, AMBP, and SERPINC1 were further verified by immunohistochemical analyses. In conclusion, these identified genes represent potential biomarkers for ICC and represent potential targets for elucidating the molecular mechanisms that are associated with ICC. 2006 Elsevier Inc. All rights reserved. Keywords: Expression profiling; ESTs frequency; Intrahepatic cholangiocarcinoma; Liver

Mortality rates from intrahepatic cholangiocarcinoma (ICC) have risen steeply and steadily over the past 30 years and, since the mid 1990s, more deaths have been recorded annually in England and Wales as being due to this type of tumor than to hepatocellular carcinoma [1]. ICC also occurs frequently in Southeast Asia and the mortality rate has been reported to be on the increase [2]. The cause of this increase is unknown and does not appear to be explainable simply by improvements in diagnosis or changes in record-keeping procedures [1]. The average survival time of ICC patients who underwent curative resection

*

Corresponding author. Fax: +82 42 879 8119. E-mail address: [email protected] (N.-S. Kim).

0006-291X/$ - see front matter 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2006.04.175

was just 2 years and, for non-curative resection, this period was no more than 3 months [3]. Studies of approximately equal numbers of men and women with ICC development indicated that the ICC may have unique pathological molecular pathways compared to the hepatocellular carcinomas (HCC) that showed a prevalence for males but not females [4]. In addition, the biliary epithelium occupies only 3–5% of the total liver nuclear population but has an incidence of about 10% in primary liver cancer [5]. This indicates that the biliary epithelium has a greater potential to develop into cancer than hepatocytes. Moreover, ICC has been described as being difficult to distinguish from HCC and exhibits a pseudoglandular or poorly differentiated morphology [6]. Although, it has been reported that ICC is associated with

A.-G. Wang et al. / Biochemical and Biophysical Research Communications 345 (2006) 1022–1032

the inactivation of certain tumor suppressor genes such as p53, APC, smad-4 as well as mutations in oncogenes such as K-ras, c-myc, and c-erb-2 [6,7], the mechanisms related to the overall process remain largely unknown. Only two published reports using a microarray to date have documented the comprehensive analysis of gene expression among cholangiocarcinoma in general [8,9]. This is largely due to the difficulty of obtaining biliary epithelium cells, that represent 3–5% of the total liver nuclear population but only about 0.1% of the parenchymal volume, as a control. In addition, the expression profiles reported in these two studies were not completely consistent with each other. These differences may be due to different sources of samples (intrahepatic cholangiocarcinomas [8], biliary carcinomas, and biliary cancer cell lines [9]) and investigating systems (cDNA microarray [8]; Affymetrix U133A gene expression microarrays [9]). Thus, these data reflect the different aspects of cholangiocarcinoma cancers. The expressed sequence tags (ESTs) generated by the single-pass sequencing of cDNA clones that are randomly selected from cDNA libraries have been used to identify novel genes [10]. ESTs have also been found to be useful for the differential and quantitative analysis of expression patterns and for the evaluation of gene expression profiles in a specific tissue [11]. In addition, the analysis using EST frequency has an advantage of selecting the candidate genes without bias in comparison with microarray analysis for only spotted genes. In a previous study, using ESTs, we identified some novel genes that offered valuable information related to the development of gastric cancer [12,13]. In order to identify potential ICC markers for diagnosis and signaling pathways related to the development of ICC, we focused on identifying differences in expression profiles between the normal liver and ICC tissues, because ICC is contained in liver tissue and is suggested to arise from the same stem cells as hepatocellular carcinoma (HCC) [14]. Nine cDNA libraries including full-length enriched cDNA libraries were constructed from ICC cell lines and normal liver tissues. Using EST frequency data obtained from liver EST data, the expression profiles of the ICC cell lines and tissues were analyzed. These data should be useful for the development of diagnostic and therapeutic strategies for treating the disease. Materials and methods Cell culture, tissue samples, and RNA preparation. Four different cell lines established from Korean intrahepatic cholangiocarcinoma (ICC) patients, Cho-CK, Choi-CK, SCK, and CK-K1 were obtained from the Chonbuk National University Medical School and Hospital, Korea [2]. The ICC cell lines reacted with neither HBV nor HCV antibodies. The cell lines were maintained at 37 C in humidified air containing 5% CO2 using RPMI1640 medium (Gibco BRL, Gaithersburg, MD) supplemented with 10% fetal bovine serum. Four normal liver tissues that were used in the construction of the cDNA libraries, N670205, N779227, N800102, and N803806, were obtained from necropsy in Eulji University School of Medicine, Korea. Five cases of tissues with pairs of ICC tumors and adjacent liver cirrhosis tissues containing the bile duct were obtained from

1023

the Catholic University of Korea College of Medicine, Korea. All patients participating in the study gave informed consent before surgery. Cases of ICC were characterized histologically by proliferating glands or tubules in intrahepatic liver tissues. The tissues were immediately frozen in liquid nitrogen. Total RNAs were extracted from the cultured cells and tissues using a commercially available RNA isolation kit (Qiagen, Hilden, Germany) following procedures recommended by the manufacturer. Construction of cDNA library and DNA sequencing. The full-length cDNA library was constructed using an improved capping method with the pCNS vector [15]. Plasmid DNAs were extracted from clones that were randomly selected in the constructed cDNA libraries using a MWG plasmidprep 96 (MWG Biotech., Ebersberg, Germany). Sequencing of the DNAs was performed using previously described procedures [13]. Bioinformatic analysis of ESTs. Analysis of the collected ESTs with the bioinformatic tool was performed according to the same previously described procedures [13]. The annotation of ‘‘high quality’’ liver ESTs was carried out using the human mRNA subset extracted from the GenBank database and the UniGene database (Hs.seq.all, Build #163) for similarity comparisons using BLASTN. For protein similarity assessment, a comparison was performed against the non-redundant protein database using BLASTX. Gene expression analysis. The frequency of each gene was analyzed by dividing the number of ESTs of a gene by the number of total clones merged into the UniGene database Build #163 in each library. Genes that were abundantly expressed in each cDNA library were selected and listed from among ESTs showing an expression frequency of >0.1%. Significant differences in gene expression between the datasets were calculated using a previously described method [16]. Analysis of expressional differences between the normal pool and the cancer pool was performed at a cut-off probability of 0.999. The gene list was sorted according to gene frequency in the pool of the over-expressing gene. The Gene Ontology (GO) database (http://www.geneontology.org/ #godatabase) was used to classify the gene profiles into the gene ontology. Each UniGene cluster was mapped to LocusLink (as of April 21, 2004) and then assigned with GO ids, as in the LocusLink database. Semi-quantitative RT-PCR. Reverse transcription (RT) was performed with 3 lg of the isolated total RNAs using the same procedures as previously described [13]. To quantify the amounts of the 1st cDNAs, the RT solution, mixed with a human B2M competitor DNA, was used as template for PCR. The PCR was performed according to the method of Kim et al. [13], the conditions for which were 1 cycle for 2 min at 94 C, and 25 cycles of 94 C for 30 s, 55 C for 1 min, and 72 C for 1 min with B2M primers (Table 1). The concentration of the 1st cDNA of each sample was adjusted based on the amounts of PCR product of B2M DNA. To validate the expression level of the selected genes, PCR was performed using adjusted 1st cDNAs as templates and a specific primer set for each gene (Table 1) with the same conditions as were used in the above PCR. The transcriptional activity of each gene was calculated relative to the average amount of target gene in normal tissues for the up-regulated or down-regulated genes in the cancer pool, and was then expressed as the relative fold expression change (log base 2), after normalization against a B2M standard. On the other hand, the transcript levels of target genes in five pairs of ICC tumor and nontumor liver cirrhosis tissues are presented as the relative fold expression change (log base 2) to the average amount of target gene in normal tissues after normalization against B2M. Immunohistochemistry. Immunohistochemical staining of ANXA1, ANXA2, AMBP, and SERPINC1 in ICC tissues was performed using the same previously described procedures [12]. The primary antibodies used were the mouse monoclonal antibody against ANXA1 (BD Biosciences Pharmingen, San Diego, CA) at a 1:200 dilution, the goat polyclonal antibody against ANXA2 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) at a 1:100 dilution, the mouse monoclonal anti-AMBP (Abcam Ltd., Cambridge, UK) at a 1:200 dilution, and the mouse monoclonal antibody against SERINC1 (Sigma–Aldrich, Inc., St. Louis, USA) at a 1:100 dilution.

1024


Table 1 Primer sequences for semiquantitative RT-PCR Genes

Forward (5 0 fi 3 0 )

Reverse (5 0 fi 3 0 )

B2M ANXA2 LDHA RPL4 ANXA1 RPLP0 GNB2L1 RPL9 HSPCA ENO1 PKM2 VIM HSPCB PGK1 RPL13A TM4SF1 ALB AMBP GC A1BG SERPINC1 FGA ADH1C HPD KNG HPX ADH1B CD14 SERPINA6 C1S ACY1

CTCGCTCCGTGGCCTTAG ATGTCTACTGTTCACGAAAT ATGGCAACTCTAAAGGATCA GAGCTGGCAAAGGCAAA ATGGCAATGGTATCAGAATTCC GTTGCTGGCCAATAAGGT AACCACATTGGCCACACA ATGAAGACTATTCTCAGCAATC ATGCCTGAGGAAACCCAGA GACTTGGCTGGCAACTCTG ATGTCGAAGCCCCATAGTGA ATG CGG GAG CTG CGC CGG CAG GTGACCAGCACCTACGGC ATGTCGCTTTCTAACAAGCTGA ATGGCGGAGGTGCAGGTC ATGTGCTATGGGAAGTGTGC TCAAAGCCTTGGTGTTGATTGCC AGTGGTACAACCTGGCCATC ATGAAGAGGGTCCTGGTACTAC ATGTCCATGCTCGTGGTCTT ATGTATTCCAATGTGATAGGAA ATGAAAGGGTTGATTGATGAAGTCA ATGGGCTGTAAAGCAGCTGGAGC ATGGGCTTTGAACCTCTAGCCTAC GCTCTGAAGAAATATAACAGTCAAA TTCCGTCAAGGTCACAACAGTGTC ATGGCACCAGGAGGTTCACCTGC GCTCGCCGAGCTGCAGCAGTGG ATGCTGTCCCTGGGCACCTGTG ATGAAGAATTGCGGAGTTAATTGCA ATGAAGTGCGTCAGCATCCAGTAC

CAAATGCGGCATCTTCAA GCTCCTGGTTGGTTCTGG GCAACTTGCAGTTCGGGC TCCGGCGCATGGTCTTT CAGTTCCAAGGCCCTTCA GCCAAGAAGGCCTTGACC TGCCAATGGTCACCTGC TGAACAAGCAACACCTGGTC TCTGCACCAGCCTGCAA GGTCATCGGGAGACTTGAA TCCACCTCTGCAGTGCC GTC AGC AAA CTT GGA TTT GTA CCA ATCGACTTCTTCCATGCGAG GCGGAGGTTCTCCAGCA GACCAGGAGTCCGTGGGT CAATGTGCTTGGGTTCAGTG ATGTCTCTTCATTGTCATGAAAAG ACAGCCCTCCGGACTCTC TCATTTGTGGGTTCCACGTA TCAGGCACCTCCAGAAACTC TCAGTTGCTGGAGGGTGTC CTCTGACAGGGCGAGATTTAGCA AGCCATAAAGTCAGCCACAAGTTT CTCCTGATCAGGCTGGTTTCCCA ATGCACACAGCCGAGGCAGTCG ATGGCCAGTCCCATTCCTGTGTC TTAGCACTTCCTGGATGGGTTTCT ATGCATCTCGGAGCGCTAGGGTT CATGGGCACCTTCACCACAGTTG TTGCCCTGTTAGATCAGTTTGGAA CTCTAGCTTAGTCAGGTTCACGG

Results Large-scale ESTs collection from human ICC cells and normal liver tissues Nine full-length cDNA libraries were constructed from four ICC cell lines and four normal liver tissues. A total of 19,991 clones were randomly selected from these 9 libraries and used for 5 0 -end single-pass sequencing. The sequences

obtained were subjected to quality control procedures, namely, trimming of the vector region and the removal of low-quality or short (less than 100 bp) sequences. Finally, 17,984 high-quality ESTs with an average length of 500 bp were collected (Table 2). After screening out the ESTs derived from mitochondrial DNAs, ribosomal DNAs, and human repetitive sequences, the remaining sequences were submitted to the NCBI dbEST database. When all of our ESTs were annotated by coalescing into human UniGene

Table 2 Summary of cDNA libraries Source

Library

Vector

Reads

Unigene 163a Clones

ICC cell lines Cho-CK

Clusters

SCK Choi-CK CK-K1

L6Cho-CK L10Cho-CK L8SCK L14Choi-CK L15CK-K1

pCNS-D2 pT7T3-Pac pT7T3-Pac pT7T3-Pac pCNS-D2

2631 2512 3324 2260 4615

2518 1993 2905 1982 4404

1258 1241 1371 1019 1979

Normal tissue N800102 N670205 N779227 N803806

L7N800102 L17N670205 L19N779227 L20N803806

pCNS-D2 pT7T3-Pac pCNS-D2 pCNS-D2

1466 845 1254 1084

1349 589 1209 1035

503 328 373 331

19991

17984

5211

Total a

Number of clones and clusters in NCBI Unigene Build 163 contributed by our EST sequences.


clusters (Build 163), they were assembled into 5211 clusters. All of them were ESTs that coded for a known gene having an identity of at least 95% with human Refseq or mRNA. These annotated results were used in analyses of the subsequent expression profiles. Identification of genes differentially expressed in ICC cell lines vs. normal liver tissues To identify the candidate genes related to human ICC, we selected libraries with more than 1,000 EST sequences and divided the constructed libraries into two groups, i.e., a ‘normal pool’ and a ‘cancer pool’. The normal pool was comprised of three libraries of normal tissues and the cancer pool of 5 libraries of cancer cells. We then selected 137 genes that had a frequency of over 0.1% on average in each pool, and which showed a significant difference (P > 0.999) between the two pools. We found 53 up-regulated genes and 84 down-regulated genes specifically in the cancer pool, as shown in Table 3. Among the up-regulated genes in the cancer pool, significant differences were mainly observed in the gene groups associated with the positive regulation of cell proliferation (GNB2L1, LAMR1, AKR1C3, NAP1L1, and PTMA), heat shock proteins (HSPCA, HSPCB, and HSPA8), the glycolysis pathway (LDHA, ENO1, PKM2, PGK1, GAPD, and LDHB), protein biosynthesis (EEF1A1, RPL4, RPL3, RPLP0, RPL9, RPL6, RPL7, RPS3A, RPL13A, RPS3, RPS11, RPL5, RPS24, RPS20, RPL23, RPL23A, RPS4X, EEF1G, RPS2, E1F4A1, RPL7A, and RPS6), cell structure formation (K-ALPHA-1 and OK/SW-cl.56), RNA processing (RBM3 and HNRPC), cytoskeletal development (ANXA2, ANXA1, and TMSB10), protein folding (PPIA), transport (NPM1), immune response (IFI27), metabolism (AKR1C1), and morphogenesis (KRT18). In the case of the down-regulated genes in the cancer pool, significant differences were mainly observed in gene groups related to the negative regulation of cell proliferation (TM4SF4, AZGP1, and ATF5), proteolysis and peptidolysis (HP, HPR, F2, PROC, CPB2, ACY1, PLG, and MST1), molecular metabolism (lipid: APOH, APOA5, RODH-4, APOA1, BAAT, ACAA1, and SLC27A5; alcohol: ADH1C, ADH1B, and ADH1A; hyaluronan: ITIH1 and ITIH4; fructose: ALDOC and FBP1; amino acid: HPD and TAT; others: AGXT and LOC348158), regulation of blood pressure and coagulation (FGG, FGB, SERPINC1, FGA, KNG, and AGT), immune response (VTN, CD74, SERPINA1, SERPING1, C1S, BF, HF1, and C8B), inflammatory response (AHSG, C3, C4A, and CRP), molecular transport (electron: CYP2A6, CYP2D6, CYP3A4, CYP2A7, CYP2E1, PIPOX, and CYP2CB; others: ALB, AMBP, GC, HPX, ORM1, SERPINA6, A2M, SLC22A1, TF, GJB1, TTR, and COLEC11), fatty acid a-oxidation (HAO2 and HAO1), copper ion homeostasis (MT2A), response to external stimulus (PON1), organogenesis (DCN), and development (ID2).

1025

Verification of up- or down-regulated genes in ICC cell lines and tissues using semi-quantitative RT-PCR To verify our data more quantitatively, we randomly selected 15 up-regulated genes (ANXA2, LDHA, PRPL4, ANXA1, RPLP0, GNB2L1, RPL9, HSPCA, ENO1, PKM2, VIM, HSPCB, PGK1, RPL13A, and TM4SF1) and 15 down-regulated genes (ALB, AMBP, GC, A1BG, SERPINC1, FGA, ADH1C, HPD, KNG, HPX, ADH1B, CD14, SERPINA6, C1S, and ACY1) for use in semi-quantitative RT-PCR. The primer information for these genes is shown in Table 1. We found that all of the up-regulated genes were highly expressed in most of the ICC cell lines but were expressed at lower levels in normal liver tissues (Fig. 1A). On the other hand, all down-regulated genes were detected at very low levels in ICC cell lines but had high expression levels in normal tissues compared to ICC cell lines (Fig. 1B). This indicates that transcriptional activity by RT-PCR was consistent with the EST frequency data for all 30 genes. To verify whether the EST results from ICC cell lines were also true for ICC tissues, genes that had been randomly selected for RT-PCR in Fig. 1 were further confirmed in ICC tissues. Normal and paired peri-tumoral tissues were used as controls. As shown in Fig. 2A and B, 13 out of 15 up-regulated genes (86.7%) and 15 out of the 15 down-regulated genes (100%) were largely consistent to the results of Fig. 1. The up-regulated genes were overexpressed and the down-regulated genes were expressed at lower levels in ICC tissues compared to the normal and peri-tumoral liver tissues. The difference in expression in up-regulated genes was observed to be higher in normal tissues than in peri-tumoral liver tissues. It may be due to the used peri-tumoral tissues which were diagnosed as cirrhosis and reflects the difference between normal liver tissues and cirrhosis tissues. These results indicate that our EST data from the ICC cell lines were firmly consistent with the results for the ICC tissues. Verification of protein levels for selected genes using immunohistochemistry To verify the protein levels for genes that were confirmed by semi-quantitative RT-PCR, an immunohistochemical analysis of 12 human ICC tissues and 4 normal liver tissues was performed. ANXA1, ANXA2, ENO1, and HSPCA for the up-regulated genes, and AMBP and SERPINC1 for the down-regulated genes were selected as target proteins. As shown in Fig. 3, the ANXA1 and ANXA2 proteins were positively expressed in the ICC region with a frequency of 83.3% (10/12) and 100% (12/12), respectively, compared with the peri-tumoral region (Fig. 3B and D) and normal liver tissue (Fig. 3A and C), and were distributed mainly in the membrane and cytosol regions. In addition, the ANXA1 and ANXA2 proteins were detected weakly or not at all in the normal bile duct epithelium (data not

1026


Table 3 Genes showing significant difference (P > 0.99) in expression between normal liver tissue and cholangiocarcinoma cell line


1027

Fig. 1. Semi-quantitative RT-PCR for genes that were randomly selected from the panels of up-regulated (A, a–o) or down-regulated (B, a–o) genes in the ICC pool based on EST frequency in ICC cell lines. Total RNAs were extracted from normal tissues (L17, N779227; L20, N803806) and ICC cell lines (L1, Cho-CK; L2, SCK; L3, Choi-CK; L4, CK-K1) and used as templates for semi-quantitative RT-PCR (for details, see Materials and methods).

1028


Fig. 2. Semi-quantitative RT-PCR of ICC tissues for genes, that had been randomly selected and confirmed in ICC cell lines. (A) (a–m), up-regulated genes in the ICC pool; (B) (a–o), down-regulated genes in the ICC pool. Total RNAs were extracted from normal tissues (L17, N779227; L20, N803806) and 5 ICC tissues and paired peri-tumoral tissues, and were used as templates for semi-quantitative RT-PCR (for details, see Materials and methods).


1029

Fig. 3. Immunohistochemical staining of ANXA1, ANXA2, AMBP, and SERPINC1 in normal liver tissues (A, C, E, and G) and ICC tissues (B, D, F, and H). ANXA1 and ANXA2 positive cells were clustered within ICC tissues (B,D), but not detected or very low expressed in stroma (B,D) or normal liver cells (A,C). AMBP and SERPINC1 positive cells were mainly detected in the cytosol of hepatocytes in normal liver tissues (E,G) and peri-tumoral tissues (F,H) but not in ICC tissues (E,F). (A) and (B), ANXA1; (C) and (D), ANXA2; (E) and (F), AMBP; (G) and (H), SERPINC1. T, ICC tissue; N, peri-tumoral tissue. Magnifications: 200· (A, B, C, E, F, G, and H); 400· (D).

shown). The AMBP (12/12, 100%) and SERPINC1 (11/12, 91.7%) secretion proteins were negatively expressed in the ICC region (T region of Fig. 3F and H) compared with normal liver tissue (Fig. 3E and G). The normal hepato-

cytes showed granular expression patterns for the AMBP and SERPINC1 in the cytoplasmic areas. On the other hand, ENO1 and HSPCA proteins were not detected in either normal liver or tumor tissues in our study. This is

1030


thought to be due to the low reactivity of the antibodies against ENO1 and HSPCA. These results indicate that the mRNA levels for the up-regulated or down-regulated target genes were largely consistent with protein levels. Discussion Intrahepatic cholangiocarcinoma is the second most common type of primary hepatobiliary cancer. Its increasing incidence in Western countries and Southeast Asia makes mechanism studies related to it all the more urgent. Although some reports have disclosed one or several genes that are involved in carcinogenesis of this disease, information relative to altered gene expression in ICCs is not readily available. A global gene expression analysis of human ICC may permit the identification of potential markers for diagnosis of ICC and provide important clues for our understanding of the oncogenesis of ICC, thus leading to improvements in our ability to predict its clinical behavior. It has been reported that the specific characteristics of genetic alterations in the original tumors are maintained during the establishment of immortalized cancer cells and, as a result, there is rarely a difference in genetic changes between the cancer cell line and the original tumor [2,17]. Our results are consistent with this conclusion, in that the EST frequency data from the ICC cell lines were in general agreement with RT-PCR data for the ICC cell lines (30 out of 30 (100%) and tissues (28 out of 30 (93.3%)). The profile of gene expression for ICC showed that genes that are involved in the positive regulation of cell proliferation such as LAMR1, AKR1C3, and NAP1L1 were up-regulated in ICC. LAMR1 is well known for its role in cancer proliferation, invasion, and metastasis [18]. It has also been shown that AKR1C3 and NAP1L1 play a role in promoting cell proliferation [19,20]. The over-expression of these genes in ICC indicates their possible important functions in the pathological process of ICC and may be new targets for ICC therapy. In addition, heat shock proteins were found to be preferentially and highly expressed in ICC. Since components related to survival and apoptotic pathways are regulated by interactions with molecular chaperones such as the HSP70 and HSP90 family [21], these heat shock proteins may be a useful target region for the molecular prognosis and treatment of ICC. Moreover, the enzymes of the glycolytic pathway such as LDHA, ENO1, PKM2, and PGK1 were increased in ICC, which is in agreement with other reports of an increase in glycolysis in many types of cancer [22,23]. These glycolytic enzymes are known to be hypoxia-inducible factors in other forms of cancer [24]. This result indicates that the HIF signaling pathway might be related to the pathogenesis and progression of ICC. Many genes that are involved in protein translation were also highly expressed in ICC, which appears to be rational since cancer cells are actively growing. These genes are also well known in HCC [25]. On the other hand, many genes for proteins that participate in the metabolism of alcohols, drugs, glucose, lipids, and amino acids and in respiratory chain complexes were

down-regulated in ICC. In addition, the levels of a number of liver-specific proteins, including albumin, transferrin, coagulation factor, and complement components were decreased, as has been reported previously [26]. In addition, in comparing the HCC tissues with normal liver tissues, a similar expression profile was observed in down-regulated genes that were detected in ICC tissues (data not shown). This indicates that, when liver cells develop into cancer, the common function of hepatocytes disappears. Moreover, genes that are related to the negative regulation the cell cycle, such as TM4SF4 [27], AZGP1 [28], and ATF5 [29], were found in the down-regulated gene in cancer pool. This indicates that these genes may have important roles in suppressing the ICC development. ANXA1 and ANXA2 are calcium-dependent phospholipid-binding proteins that play a role in the regulation of cellular growth and in signal transduction pathways [30]. They have a 50% homology at the amino acid level and are highly variable in their N-terminal regions, which are known to mediate the specific functions of individual annexins. ANXA1 has been reported to be strongly up-regulated in several human cancers, including breast cancer [31] but has recently been reported to be down-regulated in head and neck cancer [32]. ANXA2 has also been reported to be increased in colorectal cancer [33] but decreased in prostate cancer [34]. These reports indicate that these proteins have different functions, such as, for example, in tumor progression or tumor suppression, depending on the specific type of cancer or the stage of the disease. The over-expression of ANXA1 in HCC has been reported but the over-expression of ANXA2 has not. However, a strong over-expression of ANXA2 in HCC can be confirmed by the data report herein (data not shown). The results revealed that both ANXA1 and ANXA2 were over-expressed not only in HCC but also in ICC. These proteins can be used as markers for ICC especially by a sinusoidal expression pattern. The biliary epithelium is an effective and substantiative hepatocyte progenitor compartment under appropriate conditions [35]. ICC and HCC are also thought to arise from the same stem cells in liver tissues [14]. In addition, liver regeneration indicates that hepatocytes and biliary epithelial cells have a practically unlimited capacity for proliferation [36]. These data indicate that, although ICC and HCC were classified as to whether they develop from the biliary duct or hepatocytes, respectively, they may have some of the same characteristics in the tumor development. Moreover, in comparing our gene expression profile between ICC and HCC cancer cell pools, 67.5% of the up-regulated genes and 100% of the down-regulated genes in HCC cancer cells were also found to be up-regulated and down-regulated genes in ICC cancer cells, respectively (data not shown). These data offered additional evidence in support of our proposal that ICC and HCC may undergo tumoral development by some of the same molecular pathways. Using an analysis of EST frequency, we examined the difference in gene expression profiles in ICC. A comparison


of our data with two previously published reports [8,9] showed that a few genes, such as LAMR1 and: PPIA and EIF4A1 (similar category of genes) were consistent with their data and only a few genes also matched each other between these two published sets of data. This indicates that a more global gene expression analysis on cholangiocarcinoma is needed to obtain more information from different directions. In conclusion, the data reported here should lead to a better understanding of the molecular mechanisms of ICC and offer the possibility of identifying potential cancer markers for ICC. Acknowledgments The authors thank Ms. Hee-Young Ahn, Ms. Mi-Young Chu, Ms. Ju-yeon Lee, and Mr. Tae Hoon Song for their excellent technical support in ESTs sequencing. This study was supported by grant from the 21C Frontier Functional Human Genome Project from the Ministry of Science and Technology of Korea. The sequence data described in this paper have been submitted to the GenBank data library under the Accession Nos. CB105131–CB164813. References [1] S.D. Taylor-Robinson, M.B. Toledano, S. Arora, T.J. Keegan, S. Hargreaves, A. Beck, S.A. Khan, P. Elliott, H.C. Thomas, Increase in mortality rates from intrahepatic cholangiocarcinoma in England and Wales 1968–1998, Gut 48 (2001) 816–820. [2] D.G. Kim, S.Y. Park, K.R. You, G.B. Lee, H. Kim, W.S. Moon, Y.H. Chun, S.H. Park, Establishment and characterization of chromosomal aberrations in human cholangiocarcinoma cell lines by cross-species color banding, Genes Chromosomes Cancer 30 (2001) 48–56. [3] H.J. Kim, S.S. Yun, K.H. Jung, W.H. Kwun, J.H. Choi, Intrahepatic cholangiocarcinoma in Korea, J. Hepatobiliary Pancreat. Surg. 6 (1999) 142–148. [4] K. Tanaka, H. Sakai, M. Hashizume, T. Hirohata, Serum testosterone:estradiol ratio and the development of hepatocellular carcinoma among male cirrhotic patients, Cancer Res. 60 (2000) 5106–5110. [5] K. Shirabe, M. Shimada, N. Harimoto, K. Sugimachi, Y. Yamashita, E. Tsujita, S. Aishima, Intrahepatic cholangiocarcinoma: its mode of spreading and therapeutic modalities, Surgery 131 (2002) S159–S164. [6] S.A. Khan, B.R. Davidson, R. Goldin, S.P. Pereira, W.M. Rosenberg, S.D. Taylor-Robinson, A.V. Thillainayagam, H.C. Thomas, M.R. Thursz, H. Wasan, Guidelines for the diagnosis and treatment of cholangiocarcinoma: consensus document, Gut 51 (2002) VI1–VI9. [7] M.E. Reeves, R.P. DeMatteo, Genes and viruses in hepatobiliary neoplasia, Semin. Surg. Oncol. 19 (2000) 84–93. [8] K. Obama, K. Ura, M. Li, T. Katagiri, T. Tsunoda, A. Nomura, S. Satoh, Y. Nakamura, Y. Furukawa, Genome-wide analysis of gene expression in human intrahepatic cholangiocarcinoma, Hepatology 41 (2005) 1339–1348. [9] D.E. Hansel, A. Rahman, M. Hidalgo, P.J. Thuluvath, K.D. Lillemoe, R. Shulick, J.L. Ku, J.G. Park, K. Miyazaki, R. Ashfaq, Wistuba II, R. Varma, L. Hawthorne, J. Geradts, P. Argani, A. Maitra, Identification of novel cellular targets in biliary tract cancers using global gene expression technology, Am. J. Pathol. 163 (2003) 217–229. [10] M.D. Adams, J.M. Kelley, J.D. Gocayne, M. Dubnick, M.H. Polymeropoulos, H. Xiao, C.R. Merril, A. Wu, B. Olde, R.F. Moreno, et al., Complementary DNA sequencing: expressed sequence tags and human genome project, Science 252 (1991) 1651–1656.

1031

[11] K. Okubo, N. Hori, R. Matoba, T. Niiyama, A. Fukushima, Y. Kojima, K. Matsubara, Large scale cDNA sequencing for analysis of quantitative and qualitative aspects of gene expression, Nat. Genet. 2 (1992) 173–179. [12] J.M. Kim, H.Y. Sohn, S.Y. Yoon, J.H. Oh, J.O. Yang, J.H. Kim, K.S. Song, S.M. Rho, H.S. Yoo, Y.S. Kim, J.G. Kim, N.S. Kim, Identification of gastric cancer-related genes using a cDNA microarray containing novel expressed sequence tags expressed in gastric cancer cells, Clin. Cancer Res. 11 (2005) 473–482. [13] N.S. Kim, Y. Hahn, J.H. Oh, J.Y. Lee, K.J. Oh, J.M. Kim, H.S. Park, S. Kim, K.S. Song, S.M. Rho, H.S. Yoo, Y.S. Kim, Gene cataloging and expression profiling in human gastric cancer cells by expressed sequence tags, Genomics 83 (2004) 1024–1045. [14] S. Sell, H.A. Dunsford, Evidence for the stem cell origin of hepatocellular carcinoma and cholangiocarcinoma, Am. J. Pathol. 134 (1989) 1347–1363. [15] J.H. Oh, H.Y. Sohn, J.M. Kim, Y.S. Kim, N.S. Kim, Construction of multi-purpose vectors, pCNS and pCNS-D2, are suitable for collection and functional study of large-scale cDNAs, Plasmid 51 (2004) 217–226. [16] S. Audic, J.M. Claverie, The significance of digital gene expression profiles, Genome Res. 7 (1997) 986–995. [17] J.G. Park, J.H. Lee, M.S. Kang, K.J. Park, Y.M. Jeon, H.J. Lee, H.S. Kwon, H.S. Park, K.S. Yeo, K.U. Lee, et al., Characterization of cell lines established from human hepatocellular carcinoma, Int. J. Cancer 62 (1995) 276–282. [18] V. Castronovo, Laminin receptors and laminin-binding proteins during tumor invasion and metastasis, Invasion Metastasis 13 (1993) 1–30. [19] K.I. Mills, A.F. Gilkes, M. Sweeney, M.A. Choudhry, L.J. Woodgate, C.M. Bunce, G. Brown, A.K. Burnett, Identification of a retinoic acid responsive aldoketoreductase expressed in HL60 leukaemic cells, FEBS Lett. 440 (1998) 158–162. [20] H.U. Simon, G.B. Mills, M. Kozlowski, D. Hogg, D. Branch, Y. Ishimi, K.A. Siminovitch, Molecular characterization of hNRP, a cDNA encoding a human nucleosome-assembly-protein-I-related gene product involved in the induction of cell proliferation, Biochem. J. 297 (Pt. 2) (1994) 389–397. [21] D.D. Mosser, R.I. Morimoto, Molecular chaperones and the stress of oncogenesis, Oncogene 23 (2004) 2907–2918. [22] G. Weber, Enzymology of cancer cells (first of two parts), N. Engl. J. Med. 296 (1977) 486–492. [23] G. Weber, Enzymology of cancer cells (second of two parts), N. Engl. J. Med. 296 (1977) 541–551. [24] D.P. Naughton, Hypoxia-induced upregulation of the glycolytic enzyme glucose-6-phosphate isomerase perpetuates rheumatoid arthritis, Med. Hypotheses 60 (2003) 332–334. [25] M. Shuda, N. Kondoh, K. Tanaka, A. Ryo, T. Wakatsuki, A. Hada, N. Goseki, T. Igari, K. Hatsuse, T. Aihara, S. Horiuchi, M. Shichita, N. Yamamoto, M. Yamamoto, Enhanced expression of translation factor mRNAs in hepatocellular carcinoma, Anticancer Res. 20 (2000) 2489–2494. [26] X. Chen, S.T. Cheung, S. So, S.T. Fan, C. Barry, J. Higgins, K.M. Lai, J. Ji, S. Dudoit, I.O. Ng, M. Van De Rijn, D. Botstein, P.O. Brown, Gene expression patterns in human liver cancers, Mol. Biol. Cell 13 (2002) 1929–1939. [27] B.M. Wice, J.I. Gordon, A tetraspan membrane glycoprotein produced in the human intestinal epithelium and liver that can regulate cell density-dependent proliferation, J. Biol. Chem. 270 (1995) 21907–21918. [28] G. Lei, H. Brysk, I. Arany, S.K. Tyring, G. Srinivasan, M.M. Brysk, Characterization of zinc-alpha(2)-glycoprotein as a cell adhesion molecule that inhibits the proliferation of an oral tumor cell line, J. Cell Biochem. 75 (1999) 160–169. [29] D. Pati, M.L. Meistrich, S.E. Plon, Human Cdc34 and Rad6B ubiquitin-conjugating enzymes target repressors of cyclic AMPinduced transcription for proteolysis, Mol. Cell. Biol. 19 (1999) 5001–5013.

1032


[30] V. Gerke, S.E. Moss, Annexins: from structure to function, Physiol. Rev. 82 (2002) 331–371. [31] S.D. Pencil, M. Toth, Elevated levels of annexin I protein in vitro and in vivo in rat and human mammary adenocarcinoma, Clin. Exp. Metastasis 16 (1998) 113–121. [32] J.M. Garcia Pedrero, M.P. Fernandez, R.O. Morgan, A. Herrero Zapatero, M.V. Gonzalez, C. Suarez Nieto, J.P. Rodrigo, Annexin A1 down-regulation in head and neck cancer is associated with epithelial differentiation status, Am. J. Pathol. 164 (2004) 73–79. [33] T. Tomonaga, K. Matsushita, S. Yamaguchi, M. Oh-Ishi, Y. Kodera, T. Maeda, H. Shimada, T. Ochiai, F. Nomura, Identification of altered protein expression and post-translational modifi-

cations in primary colorectal cancer by using agarose twodimensional gel electrophoresis, Clin. Cancer Res. 10 (2004) 2007–2014. [34] A. Chetcuti, S.H. Margan, P. Russell, S. Mann, D.S. Millar, S.J. Clark, J. Rogers, D.J. Handelsman, Q. Dong, Loss of annexin II heavy and light chains in prostate cancer and its precursors, Cancer Res. 61 (2001) 6331–6334. [35] M.R. Alison, M. Golding, C.E. Sarraf, R.J. Edwards, E.N. Lalani, Liver damage in the rat induces hepatocyte stem cells from biliary epithelial cells, Gastroenterology 110 (1996) 1182–1190. [36] G.K. Michalopoulos, M.C. DeFrances, Liver regeneration, Science 276 (1997) 60–66.