Division of Cancer Genomics, National Cancer Center Research Institute, Tokyo, Japan
Address reprint requests to: Tatsuhiro Shibata, M.D., Ph.D., Division of Cancer Genomics, National Cancer Center Research Institute, 5-1-1, Tsukiji, Chuo-ku, Tokyo, 104-0045, Japan. E-mail: email@example.com; fax: +81-3-3248-2463.
Potential conflict of interest: Dr. Okusaka is on the speakers' bureau for and received grants from Novartis and Pfizer.
Supported in part by the Program for Promotion of Fundamental Studies in Health Sciences from the National Institute of Biomedical Innovation (NIBIO), Grants-in-Aid from the Ministry of Health, Labour and Welfare for the 3rd-term Comprehensive 10-year Strategy for Cancer Control, National Cancer Center Research and Development Funds (23-A8 and 23-B28) and The YASUDA Medical Foundation. National Cancer Center Biobank is supported by the National Cancer Center Research and Development Fund, Japan. GenBank registry numbers: AB821309 and AB821310.
Cholangiocarcinoma is an intractable cancer, with limited therapeutic options, in which the molecular mechanisms underlying tumor development remain poorly understood. Identification of a novel driver oncogene and applying it to targeted therapies for molecularly defined cancers might lead to improvements in the outcome of patients. We performed massively parallel whole transcriptome sequencing in eight specimens from cholangiocarcinoma patients without KRAS/BRAF/ROS1 alterations and identified two fusion kinase genes, FGFR2-AHCYL1 and FGFR2-BICC1. In reverse-transcriptase polymerase chain reaction (RT-PCR) screening, the FGFR2 fusion was detected in nine patients with cholangiocarcinoma (9/102), exclusively in the intrahepatic subtype (9/66, 13.6%), rarely in colorectal (1/149) and hepatocellular carcinoma (1/96), and none in gastric cancer (0/212). The rearrangements were mutually exclusive with KRAS/BRAF mutations. Expression of the fusion kinases in NIH3T3 cells activated MAPK and conferred anchorage-independent growth and in vivo tumorigenesis of subcutaneous transplanted cells in immune-compromised mice. This transforming ability was attributable to its kinase activity. Treatment with the fibroblast growth factor receptor (FGFR) kinase inhibitors BGJ398 and PD173074 effectively suppressed transformation. Conclusion: FGFR2 fusions occur in 13.6% of intrahepatic cholangiocarcinoma. The expression pattern of these fusions in association with sensitivity to FGFR inhibitors warrant a new molecular classification of cholangiocarcinoma and suggest a new therapeutic approach to the disease. (Hepatology 2014;59:1427-1434)
Cholangiocarcinoma (CC) is a highly malignant invasive carcinoma that arises through malignant transformation of cholangiocytes. It is an intractable tumor with poor prognosis, whose incidence and mortality rates are high in East Asia and have been rapidly increasing worldwide.[1, 2] CC can be subdivided into intrahepatic (ICC) and extrahepatic (ECC) types, which show distinct etiological and clinical features. ICC is the second most common primary hepatic malignancy after hepatocellular carcinoma, and is associated with hepatitis virus infection. Somatic mutations of KRAS and BRAF are the most common genetic alterations in CC.[3, 4] Surgical resection is the only curative treatment for CC, and no standard chemotherapy regimens have been established for inoperative cases or those showing recurrence after surgical resection.[5, 6]
A better understanding of the molecular basis of cancer would help develop targeted therapeutic agents against druggable genetic aberrations identified in cancer genomes.[7, 8] Tyrosine kinase inhibitors (TKIs) that target anaplastic lymphoma kinase (ALK) are particularly effective in the treatment of a distinct subset of lung adenocarcinoma carrying ALK fusions. FIG-ROS1, the first identified targetable fusion kinase in CC, has so far been reported in two patients. Very recently, a novel kinase fusion, FGFR2-BICC1, was detected in two CC cases. Thus, only a few cases harboring targetable fusion kinase genes have been reported, and the clinical characteristics of fusion-positive CC cases have not yet been described.
In the present study, we identified fibroblast growth factor receptor 2 (FGFR2) rearrangements including a novel FGFR2-AHCYL1 fusion using whole transcriptome high-throughput sequencing of tumor specimens, and determined the prevalence of FGFR2 rearrangements in CC. Our data indicate that FGFR2-fusions arise exclusively in ICC. In vitro studies suggest that FGFR2 fusion kinase is a promising candidate for targeted therapy in CC.
Materials and Methods
Clinical specimens of cholangiocarcinoma, gastric cancer, hepatocellular carcinoma, and colorectal cancer were provided by the National Cancer Center Biobank, Japan. Total RNA was extracted from grossly dissected, snap-frozen tissue using RNAspin (GE Healthcare, Little Chalfont, UK) according to the manufacturer's instructions, and RNA quality was examined using a Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA). The study protocol was approved by the Ethics Committee of the National Cancer Center, Tokyo, Japan.
Analysis of Whole Transcriptome Sequence Data
Complementary DNA (cDNA) libraries composed of 150-200 bp inserts were prepared from 2 μg of total RNA using the TruSeq RNA Sample Preparation Kit (Illumina, San Diego, CA). The libraries were subjected to paired-end sequencing of 50-100 bp fragments on the HiSeq2000 instrument (Illumina) according to the manufacturer's instructions. Paired-end reads were mapped to known RNA sequences in the RefSeq, Ensembl, and LincRNA databases using the Bowtie program (v. 0.12.5) as basically described previously. The detailed algorithm for fusion transcript detection is described in the Supporting Methods.
RT-PCR and Quantitative Real-Time PCR
Total RNA was reverse-transcribed to cDNA using SuperScript III (Life Technologies, Carlsbad, CA). The cDNA was subjected to PCR amplification using Ex-Taq (Takara Bio, Tokyo, Japan) with the following primers: FR2AHC-CF (GGACTCGCCAGAGATATCAACAATATAGAC) and FR2AHC-CR (GGACTGTGAGATCGAGCGAGAC) for FGFR2-AHCYL1 fusion, FR2BIC-CF2 (GTGTTAATGTGGGAGATCTTCACTTTAGG) and FR2BIC-CR2 (CATCCATCTTCAGTGTGACTCGATTG) for FGFR2-BICC1 fusion, FIG-e2CF1 (ACTGGTCAAAGTGCTGACTCTGGT) and ROS-e36CR2 (CAGCAAGAGACGCAGAGTCAGTTT) for FIG-ROS1 fusion, ACTB-S (CAAGAGATGGCCACGGCTGCT) and ACTB-A (TCCTTCTGCATCCTGTCGGCA) for β-actin. The PCR products were directly sequenced by Sanger sequencing using the BigDye terminator kit (Life Technologies). The expression of the FGFR2 transcript was assayed by quantitative real-time PCR (qPCR) using the LC480 thermal cycler (Roche, Penzberg, Germany). FGFR2 expression was normalized to β-actin expression. Primers used for qPCR are as follows: FGFR2 (Fwd-GGACCCAAAATGGGAGTTTC, Rev-ACCACTTGCCCAAAGCAA), β-actin (Fwd-CCAACCGCGAGAAGATGA, Rev-CCAGAGGCGTACAGGGATAG).
Fluorescent In Situ Hybridization
To identify FGFR2 rearrangements, break-apart fluorescent in situ hybridization (FISH) was performed on formalin-fixed, paraffin-embedded tumors using BAC clones corresponding to the 5′ (RP11-78A18) and 3′ (RP11-7P17) sequences flanking the FGFR2 gene and labeled by nick translation in green and red, respectively.
Four-micrometer-thick sections from formalin-fixed paraffin-embedded block were used for immunohistochemistry. Epitope retrieval was performed with trypsin (T7168, Sigma, St. Louis, MO) for 20 minutes at pH 7.7. The slides were then washed with phosphate-buffered saline (PBS) and incubated overnight with FGFR2 antibody at 4°C (1:500, ab10648, Abcam, Cambridge, UK). Immunoreactions were detected using the EnVision-FLEX system (DAKO, Glostrup, Denmark).
cDNA Cloning and Generation of Kinase-Deficient Mutants
The full-length FGFR2-AHCYL1 and FGFR2-BICC1 cDNAs were isolated from the corresponding tumor specimens by RT-PCR using PrimeSTAR GXL polymerase (Takara Bio) and primers FGFR2-H5F1 (ATGGTCAGCTGGGGTCGTTTCATCTGCCTGGTCG), AHCYL-H6R1 (GTATCTGTAATAATTAGGTTTGAATGGCCC), and BICC1-H6R1 (CCAGCGGCCACTGACACTAGCAATGTCTGA). EZR-ROS1 cDNA was reported previously. Each cDNA was subcloned into a pMXs vector (Cell Biolabs, San Diego, CA) to generate recombinant retrovirus expressing the fusion protein with a FLAG epitope tag. The kinase activity-deficient mutants were constructed by replacing tyrosine with phenylalanine at codons 568 and 569 in the FGFR2-AHCYL1 and FGFR2-BICC1 genes using a PrimeSTAR site-directed mutagenesis kit (Takara Bio).
Transforming Activity of FGFR2 Fusions
Mouse NIH3T3 fibroblast cells were infected with EZR-ROS1, FGFR2-AHCYL1, FGFR2-AHCYL1-KD, FGFR2-BICC1, or FGFR2-BICC1-KD-expressing retroviruses. Quantification of anchorage-independent growth was performed on day 12 in soft agar with the CytoSelect-96 kit (Cell Biolabs) in the presence or absence of FGFR inhibitors BGJ398 (#S2183, Selleck, Houston, TX) and PD173074 (#S1264, Selleck). The compound solution was added to the top layer of soft agar every 3 days.
Subcutaneous Transplantation in Immune-Compromised Mice
A total of 1 × 106 transduced NIH3T3 cells were injected subcutaneously into nude mice (BALB/c-nu/nu, CLEA Japan, Tokyo, Japan). Tumor formation was measured after 18 days. All animal procedures were performed with the approval of the Animal Ethics Committee of the National Cancer Center, Tokyo, Japan.
To analyze signaling, retrovirally transduced NIH3T3 cells were serum-starved for 2 hours, after which vehicle (DMSO), BGJ398, or PD173074 was added for a further 2 hours. The culture medium was then changed to standard medium containing 10% fetal bovine serum (FBS) for 10 minutes. Whole cell lysates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by transfer to a PVDF membrane. Western blot detection was performed with the WesternBreeze Chemiluminescent Immunodetection kit (Life Technologies) using primary antibodies against FLAG tag (#1E6, Wako Chemicals, Tokyo, Japan), phospho-FGFR1-4 (Tyr653, 654) (#AF3285, R&D Systems, Minneapolis, MN), STAT3 (#610189, BD, Becton Drive, NJ), phospho-STAT3 (Tyr705) (#9138, Cell Signaling Technology, Danvers, MA), p44/42 MAPK (#4695, Cell Signaling Technology), and phospho-p44/42 MAPK (Thr202/Tyr204) (#9106, Cell Signaling Technology), AKT1 (#2967, Cell Signaling Technology), and phospho-AKT (Ser473) (#4051, Cell Signaling Technology).
All data analyses were performed using JMP v. 8.02 (SAS Institute, Cary, NC). Fisher's exact test was used for categorical data, and the Student t-test was used for continuous data. Overall survival, measured from the date of surgery, was determined using the Kaplan-Meier method, and survival difference was compared using the log-rank test. Two-sided significance level was set at P < 0.05.
Identification of a Novel FGFR2 Fusion Gene
Whole transcriptome high-throughput sequencing of tumor specimens is one of the most effective methods for the identification of fusion oncogenes. Eight primary cholangiocarcinomas without KRAS/BRAF mutations or FIG-ROS1 fusion (Supporting Table 1) were analyzed to identify novel molecular alterations by massively parallel paired-end transcriptome sequencing. Aberrant paired reads that mapped to different transcription units were identified, and 17 potential fusion transcripts were predicted by our algorithm (Supporting Table 2). Sequence reads spanning the junctions of eight fusion candidate transcripts indicated in-frame gene fusion (Fig. 1A-C; Supporting Table 3) and were verified by direct sequencing of RT-PCR products spanning the breakpoints. Among these, fusion transcripts of the receptor kinase gene were detected as FGFR2-AHCYL1, FGFR2-BICC1, AHCYL1-FGFR2, and BICC1-FGFR2. However, two transcripts of AHCYL1-FGFR2 and BICC1-FGFR2 did not encode a functional protein of relevance to cancer, and conversely FGFR2-AHCYL1 and FGFR2-BICC1 were predicted to form chimeric proteins carrying the kinase domain of FGFR2 (Fig. 1D). Transcriptome sequencing showed a specific increase in the expression of the fused 3′ portion of AHCYL1 and BICC1 (Supporting Fig. 1A,B). Therefore, the formation of FGFR2-AHCYL1 or FGFR2-BICC1 might play important roles in cancer transformation.
From the tumor specimens, CC64 and CC73, we obtained cDNAs corresponding to FGFR2-AHCYL1 and FGFR2-BICC1 encoding 1,169 and 1,574 amino acids, respectively. The chimeric genes consisted of the in-frame fusion of the FGFR2 amino terminus (exons 1-19) and the AHCYL1 carboxyl terminus (exons 5-21) or the BICC1 carboxyl terminus (exons 3-21) (Fig. 1C,D; GenBank/DDBJ accession numbers AB821309 and AB821310). FGFR2-AHCYL1 is a novel FGFR2 fusion. AHCYL1 encodes an S-adenosyl-L-homocysteine hydrolase and inositol 1,4,5-trisphosphate binding protein, and contains a coiled-coil motif in the central domain. BICC1 encodes an RNA binding protein with a sterile alpha motif (SAM) protein-interaction and dimerization module at the carboxyl terminus. The FGFR2-AHCYL1 and FGFR2-BICC1 fusion proteins are likely to form homodimers through the coiled-coil motif of AHCYL1 and the SAM motif of BICC1, respectively. FGFR2, AHCYL1, and BICC1 mapped to chromosome 10q26.1, 1p13.2, and 10q21.1, respectively (Fig. 2A). FGFR2 and BICC1 are located on the long arm of chromosome 10 in opposite directions, suggesting that the FGFR2-BICC1 fusion is generated by intrachromosomal inversion (Supporting Fig. 1B). Gross rearrangement of the FGFR2 gene locus was verified by FISH with break-apart probes, which showed a split in the signals of the probes flanking the FGFR2 breakpoint in CC64 and CC73 tumors (Fig. 2B).
Prevalence of FGFR2 Fusions
RT-PCR and Sanger sequencing analysis of 102 cholangiocarcinoma specimens (66 ICCs and 36 ECCs) from Japanese individuals, including eight who had been subjected to whole transcriptome sequencing, identified seven FGFR2-AHCYL1-positive and two FGFR2-BICC1-positive cases (Table 1; Supporting Table 4). The nine FGFR2-fusion-positive cases were ICC type tumors (9/66, 13.6%). KRAS mutations were detected in 19 cases (19/102, 17.8%) and BRAF mutations in one (1/102, 1%); these mutations were mutually exclusive with the FGFR2 fusions (Fig. 3A; Supporting Table 4). Although two cases of FIG-ROS1 fusion (2/23, 8.7%) have been reported by other researchers in cholangiocarcinoma, we did not detect such fusion in this cohort. No significant differences in age, gender, tumor differentiation, clinical stage, and prognosis were detected between fusion-positive and -negative cases. (Table 2, Fig. 3B). Overall survival of ICC cases also showed no great distinction between the two groups. However, fusion positive cases had a propensity for hepatitis virus infection (Table 2). Expression of FGFR2 mRNA was significantly higher in fusion-positive cases than in fusion-negative ones (Supporting Fig. 2). Especially, KRAS/BRAF mutant cases showed reduced FGFR2 expression. This might afford collateral evidence of mutually exclusive relationships between FGFR2 fusion and KRAS/BRAF mutation. Immunohistological analysis revealed prominent FGFR2 protein expression at both cytoplasm and plasma membrane in fusion-positive cases (Supporting Fig. 3). We further screened 212 gastric cancers, 149 colorectal cancers, and 96 hepatocellular carcinomas by RT-PCR for the presence of these FGFR2 fusion transcripts. The FGFR2-BICC1 fusion gene was detected in one colorectal cancer (0.7%) and one hepatocellular carcinoma (1.0%). These fusion-positive non-ICC cases were also hepatitis virus-positive (Table 1).
Table 1. Clinical Features of FGFR2 Fusion Positive Cases
ICC: Intrahepatic cholangiocarcinoma
CRC: colorectal cancer
HCC: hepatocellular carcinoma
Table 2. Association Between Clinical Features and FGFR2 Fusion
Number of fusion positive case
Number of fusion negative case
Hepatitis virus positive
Hepatitis virus negative
FGFR2 Fusions Transform NIH3T3 Cells Both In Vitro and In Vivo
To assess the oncogenic activity of the FGFR2 fusion proteins, stable NIH3T3 clones expressing the retrovirally transfected wild-type fusion proteins or their kinase activity-deficient mutants (KD mutant) were established. As shown in Fig. 4A, wild-type FGFR2-AHCYL1 or FGFR2-BICC1-expressing cells showed anchorage-independent colony formation in soft agar, which was severely suppressed in KD mutant expressing cells. Subcutaneous transplantation of these clones into immunodeficient mice resulted in the formation of tumors from FGFR2-AHCYL1 and FGFR2-BICC1 expressing clones, whereas those expressing KD mutants did not form tumors (Fig. 4B).
To investigate the mechanisms by which the FGFR2 fusion drives oncogenesis, downstream FGFR signaling was analyzed in vitro (Fig. 5A; Supporting Fig. 4). The wild-type fusion expressing cells showed constitutive tyrosine phosphorylation in the activation loop of the FGFR kinase domain. FGFR2 signaling activates multiple downstream pathways, including RAS/MAPK and PI3K/AKT. Immunoblot analysis revealed that activation of MAPK, but not AKT or STAT3, was induced in clones expressing FGFR2-AHCYL1 and FGFR2-BHCC1. These results indicate that FGFR2 fusion proteins activate canonical FGFR signaling and confer anchorage-independent growth and in vivo tumorigenesis, both of which are hallmarks of cellular transformation.
FGFR2 Fusions Are Potential Therapeutic Targets in Cholangiocarcinoma
Next, we examined the sensitivity of FGFR2 fusion-driven tumor cells to two specific FGFR inhibitors, BGJ398 and PD173074, which selectively inhibit FGFR tyrosine kinase activity.[18, 19] These compounds significantly inhibited the phosphorylation of MAPK and reduced in vitro anchorage-independent colony formation to the level observed in KD mutant expressing cells (Fig. 5B).
FGFR genes are involved in multiple biological processes, ranging from cell transformation, angiogenesis, and tissue repair, to embryonic development. Activating point mutations and amplification of FGFR gene members have been explored as therapeutic targets in a wide range of tumors, including bladder, gastric, and lung cancers[20, 21]; however, amplification of FGFR genes is uncommon in ICC. Diverse fusions involving the FGFR gene family have also been reported in hematological and solid cancers[10, 11, 23, 24] and some have shown sensitivity to FGFR inhibition.
The identification of two recurrent FGFR2 fusions (FGFR2-AHCYL1 and FGFR2-BICC1) that are mutually exclusive with KRAS/BRAF mutations warrants a new molecular classification of cholangiocarcinoma and suggests a novel therapeutic approach in cholangiocarcinomas driven by these fusions. Wu et al. recently detected the FGFR2-BICC1 fusion gene in two cholangiocarcinoma cases, although its prevalence in cholangiocarcinoma has been lacking. The present study showed a high prevalence of FGFR2 fusion genes in the intrahepatic subtype of cholangiocarcinoma. Although two cases of another kinase fusion, FIG-ROS1 (2/23, 8.7%), have been reported by other researchers in CC, we did not detect such fusion in this study. As cholangiocarcinoma is a heterogeneous disease, some epidemiological or clinical specificity may be ascribable to the FIG-ROS1 fusion. However, no detailed pathological information of the patients was stated in that study. Further investigation is needed to clarify the whole picture of driver fusion genes in CC. Association between FGFR2 fusion positivity and hepatitis virus infection may suggest an involvement of the virus in the chromosomal rearrangements in CC. However, rare observation of FGFR2 fusion in hepatocellular carcinoma argues for further analysis of genetic rearrangements. Overexpression of the FGFR2 fusion protein hyperactivate one of the canonical signaling events downstream of FGFR. This contrast with other FGFR fusion proteins, FGFR1-TACC1 and FGFR3-TACC3 in glioblastoma, which fail to activate canonical downstream MAPK signaling, but induce aneuploidy and oncogenic transformation.
Based on the specific relevant genomic alterations, TKIs have been developed into effective therapies.[7, 8] We showed that small molecule FGFR inhibitors, BGJ398 and PD173074, efficiently blocked the downstream signaling and oncogenic activity of ICC-specific FGFR2 fusions. By the high-throughput cell line profiling assay, amplifications or mutations of FGFR genes in cancer cell lines have been reported to predict sensitivity to the selective pan-FGFR inhibitor BGJ398. This drug is currently in a phase I study in patients of advanced solid tumors with FGFR1/2 amplification or FGFR3 mutation (Novartis, Basel, Switzerland; ClinicalTrials.gov identifier: NCT01004224). Clinical investigations, akin to those conducted in other solid tumors with oncogenic fusion kinases, such as EML4-ALK, are warranted to examine the efficacy of FGFR inhibitors for the treatment of defined subset of cholangiocarcinoma harboring FGFR2 fusions.
We thank S. Wakai, H. Shimizu, S. Ohashi, W. Mukai, T. Urushidate, and N. Okada of the National Cancer Center for excellent technical assistance.
Sequencing and data analysis: Y.T., N.H., H.N., F.H.; Molecular biological analysis: Y.A., F.H.; Clinical and pathological analysis: T.Shirota., H.O., K.F., K.S., T.O., T.K.; Article writing: Y.A., Y.T., F.H., T.S.; Study design: Y.A., Y.T., T. Shibata.