SEARCH

SEARCH BY CITATION

Keywords:

  • Connectivity Map;
  • cholangiocarcinoma;
  • NVP-AUY922;
  • drug repurposing;
  • heat-shock protein 90

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

BACKGROUND.

Cholangiocarcinoma (CCA) is an aggressive tumor with a poor prognosis. There is no standard therapy for CCA, and novel drugs for treating refractory CCA need to be identified.

METHODS.

The authors hypothesized that, if a drug could reverse the gene expression signature of CCA, then it may inhibit the carcinogenesis of CCA and, hence, would be a potential therapeutic agent. Thus, the gene expression signatures from patients with CCA were queried using the bioinformatic method Connectivity Map, resulting in the enrichment of heat-shock protein 90 (HSP90) inhibitors with therapeutic potentials.

RESULTS.

Two HSP90 inhibitors, 17-AAG (tanespimycin) and the synthetic diarylisoxazole amide resorcinol NVP-AUY922, demonstrated potent antiproliferative activity in in vitro studies. In a thioacetamide-induced animal model, NVP-AUY922 also had antitumor activity and resulted in objective tumor regression. In addition, NVP-AUY922 reduced the expression of client oncoproteins involved in CCA oncogenesis and inhibited downstream proteins of both the phosphatidylinositol 3-kinase catalytic subunit α/v-akt murine thymoma viral oncogene homolog 1 protein kinase (PIK3/AKT) pathway and the v-Ki-ras2 Kirsten rat sarcoma viral oncogene/mitogen-activated protein kinase (KRAS/MAPK) pathway.

CONCLUSIONS.

Preclinical data from the current study suggest that NVP-AUY922 may be an effective treatment option for patients with CCA. Cancer 2013. © 2012 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Cholangiocarcinoma (CCA) is a malignant cancer arising from the neoplastic transformation of the cholangiocytes that line the intrahepatic and extrahepatic bile ducts.1, 2 CCA is the second most common primary hepatic malignancy, and recent epidemiologic studies suggest that the incidence and mortality rates of CCA, especially intrahepatic CCA, are increasing worldwide.2-4 Most patients with CCA are diagnosed with advanced disease, and curative surgical resection is possible but is sometimes difficult to perform.4, 5 Several molecular-targeted therapies have been assessed in clinical trials and produced limited median progression-free survival ranging from 1.8 months to 7 months.6, 7 However, to date, there is no standard therapy for refractory CCA, and identifying additional therapeutic drugs is urgently needed.

We previously demonstrated that, if a drug could at least partially reverse the gene expression signature of cancer cells, then it also may inhibit cancer-related pathways and, thus, treat cancer. We used the gene signatures of hepatocellular carcinoma and the Connectivity Map (CMap) tool8—a combination of computational and experimental studies—and screened antipsychotic drugs that could be used for human hepatocellular carcinoma.9 The CMap project hosts approximately 6100 gene expression profiles using Affymetrix gene chips (Affymetrix, Santa Clara, Calif) from cultured human cancer cell lines treated with bioactive small molecules and provides pattern-matching algorithms to mine these data.9 This platform-independent system uses a nonparametric, rank-based algorithm to calculate a score that indicates the degree of similarity or dissimilarity between query gene signatures and profile gene signatures. A strong negative connectivity score (dissimilarity) indicates that an agent reverses the expression of corresponding gene signatures. Thus, agents with strong negative connectivity scores may allow a transition from a particular diseased state to a more phenotypically normal state.9-13 This concept was applied in the current study to identify drugs that could reverse the gene expression signature of CCA. We collected gene signatures of intrahepatic CCA, queried the CMap database with the signatures, and identified potential therapeutic drugs. Our analysis identified heat-shock protein 90 (HSP90) inhibitors as 1 class of candidate agents.

The molecular chaperone HSP90 plays an important role in the post-translational maturation and activation of many critical oncogenic client proteins that are essential for facilitating malignant transformation and increasing the survival, growth, and invasive potential of cancer cells.13, 14 HSP90 inhibitors induce proteasome-mediated degradation of these client proteins and, thus, inhibit the growth and survival of cancer cells. Herein, we demonstrate that 2 HSP90 inhibitors, 17-AAG (tanespimycin) and the synthetic diarylisoxazole amide resorcinol NVP-AUY922, exhibit potent antitumor activity at tolerated doses in a CCA model. Our preclinical data provide a rationale to conduct clinical trials with NVP-AUY922 in patients with CCA.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Patients and Tumor Type Validation

We prospectively collected pathologically proven intrahepatic CCA specimens from 8 patients who had undergone hepatectomy at the Chang Gung Memorial Hospital. The study protocol was approved by the Institutional Review Board (IRB 94-672B) at the Chang Gung Memorial Hospital. There were 4 men and 4 women, and the median patient age was 56.7 years. The tumor types of all 8 CCA samples were validated histologically by positive immunoreactivity to the biliary cytokeratin (CK) CK7/CK19 (Dako, Carpinteria, Calif) and negative immunoreactivity to HepPar1 (Dako).

Microarray Analysis

We used 8 pairs of human CCA specimens and normal specimens in expression arrays that were manufactured for gene expression analysis by NimbleGen (Madison, Wis). The gene expression signature was based on a comparison between 8 pairs of tumor tissue and nontumor, normal liver tissue from the same individual.15, 16 The microarrays were analyzed using GeneSpring (Silicon Genetics, Redwood City, Calif) with the default settings. The robust multichip average with default settings was used for normalization and summarization of the data. T tests for paired data were conducted, and genes that had P values < .01 were selected for further analysis.

Connectivity Map Analysis

The steps involved in CMap analysis are illustrated in Figure 1. Genes that were up-regulated or down-regulated were entered into the CMap database. Only drugs that had negative scores and P values < .05 were considered for further study. The sum of drug occurrences was used to rank the drugs.

thumbnail image

Figure 1. This schematic illustrates the major steps in the current combined computational and experimental approaches for identifying heat-shock protein 90 (HSP90) inhibitors as potential therapeutic drugs in cholangiocarcinoma. Gene expression signatures of cholangiocarcinoma were used to query the Connectivity Map database, which identified HSP90 inhibitors as potential therapeutic drugs in cholangiocarcinoma. (Portions of this figure were adapted from: Lamb J, Crawford ED, Peck D, et al The Connectivity Map: using gene-expression signatures to connect small molecules, genes, and disease. Science. 2006;313:1929-19358; and Palchaudhuri R, Hergenrother PJ. Transcript profiling and RNA interference as tools to identify small molecule mechanisms and therapeutic potential. ACS Chem Biol. 2011;6:21-33.10)

Download figure to PowerPoint

Cell Lines

Two intrahepatic CCA cell lines were obtained from the following sources: HuCCT1 (Japanese Collection of Research Bioresources, Osaka, Japan) and CGCCA (Chang Gung Memorial Hospital, Taoyuan, Taiwan).17, 18 HuCCT1 and CGCCA cells were routinely cultured in RPMI1640 and Dulbecco modified Eagle medium (Gibco, Grand Island, NY), respectively, supplemented with 10% heat-inactivated fetal bovine serum, 100 μg/mL streptomycin, 100 μg/mL penicillin, and 2 mM l-glutamine in a humidified atmosphere containing 5% CO2 at 37°C.

Chemicals

The drug 17-allylamino-demethoxy geldanamycin (17-AAG) was purchased from Sigma (St. Louis, Mo). A stock solution was prepared in 100% dimethyl sulfoxide (DMSO) and stored at −20°C. The drug was diluted in fresh media before each experiment. NVP-AUY922 was provided by Novartis (Basel, Switzerland). For in vitro experiments, stock solutions of NVP-AUY922 were prepared in 100% DMSO (10 mM) and stored at −20°C. For subcutaneous injections, the free base of NVP-AUY922 was formulated in 60 mM lactic acid or 2.5% ethanol, 20% 50 mM tartaric acid, and 77.5% of 5% glucose in water containing 1% Tween-80 (volume/volume). An optimized NVP-AUY922 salt with high solubility in aqueous solution was formulated in 5% glucose in water for subcutaneous administration and delivered in a volume of 10 mL/kg.

Cell Cycle Analysis

Cells were plated on 6-well plates, incubated for 24 hours, and then treated with DMSO, 17-AAG, or NVP-AUY922. For flow cytometry, cells were trypsinized and fixed in 70% ethanol at −20°C, washed, and incubated with 10 mg/mL RNase A (Sigma) for 15 minutes at 37°C. Next, the cells were stained with 200 μg/mL propidium iodide (Sigma) for 1 hour at room temperature. Cells were evaluated using a FACSCalibur machine (Becton Dickinson, Franklin Lakes, NJ), and the data were analyzed using CellQuest software (Becton Dickinson) for modeling cell cycle distribution. Experiments were performed in triplicate, and data are expressed as the mean ± standard deviation.

Tetrazolium Salt 3-(4, 5-Dimethylthiazol-2-yl)-2, 5-Diphenyltetrazolium Bromide Cell Viability Assay

The viability of the exposed cells was determined using the TACS tetrazolium salt 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) cell proliferation assay kit (Trevigen, Gaithersburg, Md) according to the manufacturer's instructions. MTT is used to determine cell viability in cell proliferation and cytotoxicity assays. Briefly, the cells were seeded at a concentration of 1500 cells per well (HuCCT1 cells) or 900 cells per well (CGCCA cells) in 50 μL of culture medium into 96-well microplates. At 24 hours postseeding, the cells were treated with DMSO, 17-AAG, or NVP-AUY922. The cells were exposed to different concentrations of the drugs for 72 hours. Subsequently, the cells were incubated in medium containing MTT for 2 hours. The optical density at 450 nm was measured using a microplate reader (Spectral Max250; Molecular Devices, Sunnyvale, Calif).

Western Blot Analysis

Whole cell lysates from CCA cell lines were obtained using a Pierce radioimmunoprecipitation assay buffer (Thermo Scientific, Rockford, Ill). Protein samples were separated on 8% to 12% gradient dodecyl sulfate-polyacrylamide gels and transferred to Immobilion-P membranes (Millipore, Billerica, Mass). Antigen-antibody complexes were detected using an electrochemiluminescence blotting analysis system (Millipore). The following primary antibodies were used: Hsp70, v-akt murine thymoma viral oncogene homolog 1 protein kinase (AKT), phosphorylated AKT (p-AKT), cleaved poly(ADP-ribose) polymerase (PARP), caspase 3, cell division cycle 25 homolog C (Cdc25c), platelet-derived growth factor receptor beta (PDGFR)-β, phosphorylated I kappa B-alpha (p-IκBα), extracellular signal-regulated kinase (ERK), and p-ERK all from Cell Signaling Technology (Danvers, Mass); β-actin, epidermal growth factor receptor (EGFR), and insulin-like growth factor 1 receptor (IGF1R) from Abcam (Cambridge, United Kingdom); vascular endothelial growth factor receptor 2 (VEGFR2) from Millipore; met proto-oncogene (hepatocyte growth factor receptor) (c-MET) from Santa Cruz Biotechnology (Santa Cruz, Calif); v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog (c-KIT) from Dako; and v-raf murine sarcoma viral oncogene homolog B1 (B-raf) from GeneTex, Inc. (San Antonio, Tex).

Animal Studies

All animal studies were approved by the experimental animal ethics committee of the Chang Gung Memorial Hospital. All investigations conformed to the US National Institutes of Health Guide for the Care and Use of Laboratory Animals.19 Twelve adult male Sprague-Dawley rats (weight, 310 ± 14 g) were used in the experiments. Animals were equally divided into a control group and an experimental group. The rats were housed in an animal room with a 12-hour:12-hour light-dark cycle (light from 08:00 AM to 08:00 PM) at an ambient temperature of 22 ± 1°C. Food and water were provided ad libitum. The rats were administered 300 mg thioacetamide (TAA) in 1 L of drinking water daily for up to 24 weeks. In the experimental group, NVP-AUY922 (12.5 mg/kg, subcutaneously) was administered once daily 5 days per week over a 2-week period; specifically, the from the 21st week to the 22nd week. Rats in the control group received a subcutaneous injection of buffer according to the same schedule.

Positron Emission Tomography

To evaluate the changes in glycolysis in live animals with liver tumors, we conducted 2-deoxy-2-[F-18]fluoro-D-glucose (FDG)-positron emission tomography (PET) studies in rats at the molecular imaging center of the Chang Gung Memorial Hospital. In total, 12 rats were received TAA and were subjected to serial PET scanning in weeks 20, 22, and 24 using the Inveon system (Siemens Medical Solutions USA, Inc., Knoxville, Tenn). Equal numbers of animals were assigned to the control and treatment groups according to the baseline PET findings. In other words, the control and treatment groups possessed similar PET-positive rates. The details of radioligand preparation, scanning protocols, and the determination of optimal scanning time have been described previously by our group.20 Briefly, animals were fasted overnight before the scan. At 90 minutes after intravenous FDG injection, 30-minute static scans were obtained for all animals. All imaging studies were performed using a temperature-controlled (set to 37°C) and anesthesia gas-controlled (2% isoflurane in 100% oxygen) imaging bed (Minerve System; Bioscan, Wash, DC). PET images were reconstructed using the 2-dimensional-ordered subset expectation-maximization method (4 iterations and 16 subsets) without attenuation or scatter corrections. All imaging data were processed using the PMOD image-analysis workstation (PMOD Technologies Ltd., Zurich, Switzerland). The largest liver tumor was identified by careful investigation of all 3 image sets for each rat. FDG uptake into the biggest liver tumor and into apparently normal liver tissue was quantified by calculating the standardized uptake value (SUV) according to the following formula:

  • equation image

These values were calculated according to the recommendations of the European Organization for Research and Treatment of Cancer.21 The tumor regions of interest were determined using transverse images of the selected tumors and measuring the greatest dimension. Normal liver regions of interest also were determined using the same transverse images. The mean SUV (SUVmean) of the normal liver and tumor tissue was determined, and the tumor-to-liver radioactivity ratio was calculated for comparison.

Statistical Analysis

Two-sided Student t tests were used for statistical comparisons. All P values < .05 were considered statistically significant.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Gene Signatures of Cholangiocarcinoma and Connectivity Map Analysis

A t test with a P value threshold of .01 was used to analyze the microarray data. Based on this cutoff value, 58 up-regulated genes and 47 down-regulated genes were identified. Because the CMap platform was based on Affymetrix arrays, a conversion step was required to match the NimbleGen probe sets into their corresponding Affymetrix U133A probe sets (Table 1). Probe sets with the same gene symbol annotations from both platforms were considered correct matches. Of the genes that had altered expression, 11 up-regulated genes and 2 down-regulated genes had no matches to the U133A platform. Because a gene may match to multiple Affymetrix probe sets, 47 up-regulated genes and 45 down-regulated genes were matched to 73 up-regulated probe sets and 73 down-regulated probe sets, respectively. The up-regulated and down-regulated probe sets were queried against CMap, and the results were filtered at P values < .05. The top 10 drugs identified by CMap analysis are listed in Table 2. Three of the top 10 drugs, 17-AAG (tanespimycin), geldanamycin, and alvespimycin, are HSP90 inhibitors, which suggested that HSP90 may be a good drug target for CCA (Fig. 2). We wanted to further validate the antitumor effects of HSP90 inhibitors in CCA, and we chose 2 available HSP90 inhibitors, 17-AAG and NVP-AUY922, for additional study. NVP-AUY922 is a novel HSP90 inhibitor; however, it is not included in the CMap database.

thumbnail image

Figure 2. Heat-shock protein 90 (HSP90) inhibitor have a strong, negative correlation with cholangiocarcinoma signatures. The “bar view” is constructed from 6100 horizontal lines, each representing an individual treatment instance, ranked by their correlation with the query (cmap indicates Connectivity Map). The presence of tanespimycin (17-AAG), geldanamycin, and alvespimycin in the data set is indicated in black. Colors applied to the remaining entries reflect their scores (green, positive; gray, null; red, negative). The rank, drug name, concentration, cell line, and score for each of the selected HSP90 inhibitors are indicated.

Download figure to PowerPoint

Table 1. The 58 Up-Regulated and the 47 Down-Regulated Genes Identified by Analyzing Data From 8 Patients With Cholangiocarcinoma
Up-Regulated GenesDown-Regulated Genes
Gene NameDefinitionGene NameDefinition
TESCTescalcinAZGP1Alpha-2-glycoprotein 1, zinc-binding
C19orf33Chromosome 19 open reading frame 33MST1Macrophage stimulating 1 (hepatocyte growth factor-like)
MUC1Mucin 1, cell surface associatedUGT2B10Uridine diphosphate glucuronosyltransferase 2 family, polypeptide B10
SFNStratifinPCK2Phosphoenolpyruvate carboxykinase 2 (mitochondrial)
KRT17Keratin 17DCXRDicarbonyl/L-xylulose reductase
S100A14S100 calcium binding protein A14VTNVitronectin
THBS2Thrombospondin 2KHKKetohexokinase (fructokinase)
SLC16A3Solute carrier family 16, member 3SERPINF2Serpin peptidase inhibitor, clade F (alpha-2 antiplasmin, pigment epithelium derived factor), member 2
LAMB3Lamiknin, beta 3UGT2B7Uridine diphosphate glucuronosyltransferase 2 family, polypeptide B7
CAPGCapping protein (actin filament), gelsolin-likeAPOC1Apolipoprotein C-1
KRT19Keratin 19ORM1Orosomucoid 1
CTHRC1Collagen triple helix repeat containing 1LOC123876Gene encoding hypothetical protein LOC123876
COMPCartilage oligomeric matrix proteinAMBPAlpha-1-microglobulin/bikunin precursor
PKM2Pyruvate kinase, muscleAPOC4Apolipoprotein C-4
MMP11Matrix metallopeptidase 11ACSM2Acyl-coenzyme A synthetase medium-chain family member 2
THY1Thymus-1 cell surface antigenF9Coagulation factor IX
COL5A1Collagen, type V, alpha 1PON1Paraoxonase 1
MUC13Mucin 13FGGFibrinogen gamma chain
PDZK1IP1PDZK1 interacting protein 1FGL1Fibrinogen-like 1
S100A11S100 calcium binding protein A11HPHaptoglobin
USH1CUsher syndrome 1C (autosomal recessive, severe)HPXHemopexin
TPM2Tropomyosin 2 (beta)APOBApoplipoprotein B (including Ag[x] antigen)
CRIP1Cysteine-rich protein 1 (intestinal)CYP2A6Cytochrome P450, family 2, subfamily A, polypeptide 6
LOC389667Similar to tropomyosin 4CYP2A7Cytochrome P450, family 2, subfamily A, polypeptide 7
CD248CD248 molecule, endosialinF2Coagulation factor II (thrombin)
SPP1Secreted phosphorprotein 1FGBFibrinogen beta chain
DBN1Drebrin 1GSTA1Glutathione S-transferase alpha 1
NESNestinRBP4Retinol binding protein 4, plasma
SPHK1Sphingosine kinase 1ALBAlbumin
NXNNucleoredoxinHPRHaptoglobin-related protein
LOC653091Similar to cis-Golgi matrix protein GM130MT1GMetallothionein 1G
ADAM8A disintegrin and metallopeptidase domain 8APOHApolipoprotein H (beta-2-glycoproein I)
MSLNMesothelinTFTransferrin
LFNGO-fucosylpeptide 3-beta-N-acetylglucosaminyltransferaseITIH1Inter-alpha-trypsin inhibitor heavy chain 1
LOC58489Family with sequence similarity 108, member C1 (FAM108C1)APOC3Apolipoprotein C-III
ENPP2Ectonucleotide pyrophosphatase/phosphodiesterase 2SERPINC1Serpin peptidase inhibitor, clade C (antithrombin), member 1
HTRA3HrtA serine peptidase 3AHSGAlpha-2-HS-glycoprotein
PDLIM7PDZ and LIM domain 7 (enigma)ALDOBAldolase B, fructose-bisphosphate
LOC388022Hypothetical gene supported by AK131040SAA4Serum amyloid 4A, constitutive
LOC653047Similar to TCB1 domain family member 3FABP1Fatty acid binding 1, liver
JAG2Jagged 2HRGHistidine-rich glycoprotein
ABLIM2Actin binding LIM protein family, member 2APOA2Apolipoprotein A-II
TBC1D3TBC1 domain family, member 3TTRTranshyretin
TBC1D3CTBC1 domain family, member 3CAPOA1Apolipoprotein A-I
TACC3Transforming, acidic coiled-coil containing protein 3  
NCDNNeurochondrin  
RBM35AEpithelial splicing regulatory protein 1  
PAQR6Progestin and adipoQ receptor family member IV  
CCNFCyclin F  
TMEM16JTransmembrane protein 16J (anoctamin 9)  
C1orf106Chromosome 1 open reading frame 106  
LMNB2Lamin B2  
FAM83HFamily with sequence similarity 83, member H  
KLF5Kruppel-like factor 5 (intestinal)  
MAGED4Melanoma antigen family D4  
Table 2. The Top 10 Potential Therapeutic Drugs for Cholangiocarcinoma as Determined by Connectivity Map Analysis
RankDrugMechanism
  1. Abbreviations: HSP90, heat-shock protein 90; PI3K, phosphatidylinositol 3-kinase.

117-AAG (Tanespimycin)HSP90 inhibitor
2LY-294002PI3K inhibitor
3GeldanamycinHSP90 inhibitor
4AlvespimycinHSP90 inhibitor
50297417-0002BUnknown
6PuromycinAntibiotic
7PralidoximeAntidote to organophosphate pesticides and chemicals
8LomustineAnticancer chemotherapeutic drugs
9LevothyroxineSynthetic form of thyroxine (thyroid hormone)
10Prestwick-1080Unknown

Tanespimycin and NVP-AUY922 Induce Cell Growth Inhibition in Cholangiocarcinoma Cell Lines

17-AAG and NVP-AUY922 exhibited strong antiproliferative effects in both HuCCT1 cells and CGCCA cells (Fig. 3A). In the HUCCT1 and CGCCA cell lines, the 50 inhibitory concentration (IC50) values of 17-AAG were 58 nM and 47 nM, respectively, and the IC50 values of NVP-AUY922 were 40 nM and 15 nM, respectively (Fig. 3B). Cellular potency, as determined from the mean IC50 values for the 2 cell lines tested, was higher for NVP-AUY922 than for 17-AAG by a factor of 1.5 in CGCCA cells and by a factor of 3.1 in HuCCT1 cells (Fig. 3B).

thumbnail image

Figure 3. Potent cell growth inhibition induced by NVP-AUY922 and tanespimycin (17-AAG) is illustrated in cholangiocarcinoma cell lines. (A) The antiproliferative effects of 17-AAG in the CGCCA and HuCCT1 cell lines are illustrated. (B) The antiproliferative effects of NVP-AUY922 in the CGCCA and HuCCT1 cell lines are illustrated. CGCCA and HuCCT1 cells were incubated with various concentrations (1 nM, 2 nM, 4 nM, 8 nM, 16 nM, 32 nM, 64 nM, and 128 nM) of 17-AAG or NVP-AUY922 for 72 hours. Cell viability was evaluated using the tetrazolium salt 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) cell-viability assay. The data represent the mean ± standard deviation of 3 independent experiments. IC50 indicates 50% inhibitory concentration.

Download figure to PowerPoint

Tanespimycin and NVP-AUY922 induce G2/M Arrest and Apoptotic Cell Death in Cholangiocarcinoma Cells

We further demonstrated G2/M phase cell cycle arrest in 17-AAG-treated and NVP-AUY922-treated CGCCA cells (Fig. 4A). In addition, Cdc25C, a key regulator of G2/M cell division, was down-regulated in drug-treated CGCCA and HuCCT1 cells (Fig. 4A). Furthermore, 17-AAG and NVP-AUY922 induced apoptosis in the cancer cell lines. This was evident by the time-dependent and dose-dependent increase in the sub-G1 population and cleaved PARP (Fig. 4B-D). PARP cleavage resulted in the simultaneous activation of the caspase pathway, as indicated by an increase in the levels of cleaved caspase 3 (Fig. 4D).

thumbnail image

Figure 4. Tanespimycin (17-AAG) and NVP-AUY922 induce G2/M arrest and cell apoptosis in cholangiocarcinoma (CCA) cells. (A) The cell cycle profile was determined by DNA content analysis using propidium iodine (PI) staining and flow cytometry at various concentrations (0 nM, 50 nM, 100 nM, 200 nM, 500 nM, and 1000 nM). Immunoblot analysis with the primary antibody cdc25C is illustrated at representative time points. β-Actin was used as a loading control. (B) CGCCA and HuCCT1 cells were incubated with various concentrations (0 nM, 50 nM, 100 nM, 200 nM, 500 nM, and 1000 nM) of either 17-AAG or NVP-AUY922 for 72 hours. (C) CGCCA and HuCCT1 cells were treated with 5 times the 50% inhibitory concentration (IC50) of either NVP-AUY922 or 17-AAG at time zero and were harvested at the indicated time points. The number of cells in sub-G1 phase, as determined by flow cytometry, are represented as a percentage of total events. Values are depicted as the mean ± standard deviation of at least 3 independent experiments. (D) These blots illustrate immunoblot analyses of cleaved poly(ADP-ribose) polymerase (PARP) and cleaved caspase 3 with β-actin used as a loading control.

Download figure to PowerPoint

Tanespimycin and NVP-AUY922 Induce the Molecular Signature of Heat-Shock Protein 90 Inhibition in Cholangiocarcinoma Cell Lines

The data demonstrating HSP70 as a marker of the heat-shock transcription factor-1–dependent stress response are provided in Figure 5. When CGCCA and HuCCT1 cells were treated with 5 times the IC50 value of 17-AAG or NVP-AUY922, an increase in the level of HSP70 was detected after 8 hours of treatment. Western blot analyses of various HSP90 client protein expression levels in response to NVP-AUY922 and 17-AAG treatment over a 72-hour period in the CGCCA and HuCCT1 cell lines are provided in Figure 5. Both drugs depleted levels of the main growth factor receptor tyrosine kinases expressed in CCA cell lines, including the EGFR, IGF1R, c-MET, and c-KIT client proteins. In both cell lines, 17-AAG and NVP-AUY922 decreased the level of p-IκBα, although it is not a client protein.

thumbnail image

Figure 5. Western blot analysis reveals the molecular signature of heat-shock protein 90 (HSP90) inhibition induced by tanespimycin (17-AAG) and NVP-AUY922 in cholangiocarcinoma cell lines. Cell lysates from the CGCCA and HuCCT1 cell lines were treated for 72 hours with 5 times the 50% inhibitory concentration (IC50) of either 17-AAG or NVP-AUY922. β-Actin was used as a loading control. Akt indicates v-akt murine thymoma viral oncogene homolog 1; p-Akt, phosphorylated Akt; B-raf, v-raf murine sarcoma viral oncogene homolog B1; Erk, extracellular signal-regulated kinase 2 (mitogen activated protein kinase 1); p-Erk; phosphorylated Erk; VEGFR-2, vascular endothelial growth factor receptor 2; PDGFRβ, platelet-derived growth factor receptor beta; EGFR, epidermal growth factor receptor; IGF1R, insulin-like growth factor 1 receptor; c-MET, met proto-oncogene (hepatocyte growth factor receptor); c-Kit, v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog; p-IκB;agr;, phosphorylated nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor, alpha.

Download figure to PowerPoint

Tanespimycin and NVP-AUY922 Inhibit Downstream Proteins of Both the Phosphatidylinositol 3-Kinase Catalytic Subunit α/AKT Pathway and the v-Ki-ras2 Kirsten Rat Sarcoma Viral Oncogene/Mitogen-Activated Protein Kinase Pathway

The decreased expression of AKT and B-raf, 2 important HSP90 client proteins, was observed after NVP-AUY922 and 17-AAG treatment over a 72-hour period, as indicated in Figure 5. Reduced AKT and B-raf levels resulted in the simultaneous abrogation of phosphatidylinositol 3-kinase catalytic subunit α (PIK3A)/AKT pathway activity and v-Ki-ras2 Kirsten rat sarcoma viral oncogene (KRAS)/ mitogen-activated protein kinase (MAPK) pathway activity, as indicated by a decrease in p-AKT and p-ERK levels (Fig. 5).

NVP-AUY922 Induces Tumor Regression in a Thioacetamide-Treated Animal Model

In an in vivo experiment, we evaluated transverse, sagittal, and coronal PET scan views of tumor tissues from animals with TAA-induced CCA. The control and experimental rats had at least 1 FDG-avid tumor with increased SUV in the liver after treatment with TAA. Representative PET images are provided in Figure 6A,B. The tumor/liver (T/L) SUV ratios in the control and NVP-AUY922-treated groups at different time points after TAA treatment are illustrated in Figure 6C. The mean T/L SUV ratio remained high up to the 24th week in the control group. The mean T/L SUVmean ratio was significantly lower at 2 weeks post-treatment with NVP-AUY922 than that determined in the control animals (control vs NVP-AUY922: 1.81 ± 0.18 vs 1.40 ± 0.17; P < .05). The PET analysis results suggested that NVP-AUY922 treatment resulted in partial but significant suppression of tumor growth in a rat CCA model.

thumbnail image

Figure 6. (A,B) These transverse 2-deoxy-2-(F-18) fluoro-D-glucose (FDG)-positron emission tomography (PET) scans at the tumor level reveal (A) tumor progression in control rats and (B) tumor regression after treatment with NVP-AUY922 (12.5 mg/kg). The image intensity was normalized to normal liver tissue for comparison. The color scale represents the intensity of the specific uptake value ratio (SUVR) of the tumor to normal liver (arrowheads indicate cholangiocarcinomas). (C) Tumor regression is apparent in the third PET scan from the treatment group (NVP-AUY922 12.5 mg/kg; n = 6) compared with the scan from the control group (n = 6). Each point represents the mean ± standard deviation. L/T ratio indicates liver-to-tumor ratio.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

To identify new therapeutic options for the treatment of CCA, we collected CCA-related genes to query CMap and demonstrated that HSP90 inhibitors are potential therapies for human CCA. The HSP90 inhibitors 17-AAG and NVP-AUY922 have significant proapoptotic and antiproliferative effects in CCA cell lines in vitro. Moreover, NVP-AUY922 induces tumor regression in a TAA-induced CCA animal model as determined by PET analysis.

The need to identify new drugs for the treatment of CCA while ensuring cost effectiveness encouraged us to explore a drug-repurposing methodology. With bioinformatics analysis tools, we exploited CCA gene signatures and CMap8 to define associations between gene perturbations, diseases, and drug activities. This allowed us to systematically identify several potential therapeutic drugs. It is noteworthy that 3 of the 10 top ranking drugs, namely, tanespimycin, geldanamycin, and alvespimycin, were HSP90 inhibitors. This finding prompted us to investigate the role of HSP90 inhibitors in CCA. To our knowledge, this is the first study to demonstrate antitumor activity of HSP90 inhibitors in CCA.

Knowledge of the pathogenesis of tumors and its inhibition can help identify cancer-targeting agents. On the basis of this rationale, several cancer-targeting agents have been identified; for example, human epidermal growth factor receptor 2 (HER2)-blocking antibodies in breast cancer,22 EGFR monoclonal antibodies in KRAS wild-type colon cancer,23 and EGFR tyrosine kinase inhibitors in lung cancer with EGFR mutations.24 For CCA, several important tumor-associated pathways, including the KRAS/MAPK, EGFR, HER2/NEU, and IGF/PIK3 pathways, have been identified.6, 7 We observed that 17-AAG and NVP-AUY922 depleted EGFR and IGF1R, the main client proteins involved in CCA oncogenesis, and inhibited downstream proteins in both the PI3K/AKT pathway and the KRAS/MAPK pathway; treatment also efficiently inhibited their downstream proteins, p-AKT and p-ERK. Because CCA arises through the cumulative action of multiple oncogenic changes, it may acquire resistance to targeted therapies by activating compensatory pathways. Therefore, 17-AAG and NVP-AUY922, which can inhibit several client oncoproteins and their downstream pathways, may be considered therapeutic agents in CCA.

The geldanamycin analog 17-AAG (tanespimycin) and alvespimycin have exhibited tumor responses in some phase 1/2 clinical trials,25, 26 and they in are the same class of HSP90 inhibitors. Thus, we have provided the results from 17-AAG in this study for comparison. However, these compounds have some limitations, including severe adverse events, such as hepatotoxicity.26 17-AAG also is difficult to produce, and there is only limited time remaining before the drug's patent expires. Thus, the pharmaceutical company halted development of 17-AAG in 2010. In our bioinformatic analysis (Figs. 1, 2), we identified HSP90 as a potential target for CCA. To translate our findings into future clinical practice, for the current study, we chose another novel HSP90 inhibitor that is currently in development: NVP-AUY922. The synthetic diarylisoxazole amide resorcinol, NVP-AUY922, is 1 of the most potent small-molecule HSP90 inhibitors.27 The IC50 of NVP-AUY922 was lower than that of 17-AAG, indicating that NVP-AUY922 is more potent. NVP-AUY922 demonstrated antiproliferative activity in both CCA cell lines that we studied and also decreased the level of p-IκBα (Fig. 5). This suggests that the antiproliferative activity of NVP-AUY922 may be caused by inhibition of the nuclear factor κB pathway.28 NVP-AUY922 also reportedly demonstrated antiproliferative activity against other tumors in previous preclinical trials.29-31

We previously developed a TAA-induced CCA rat model that recapitulated the histologic progression of human CCA.18 The results indicated that TAA rat model may serve as a useful preclinical tool for evaluating therapeutic strategies in invasive CCA.18

In recent years, tumor response to therapeutic agents has been detected by using a combination of direct tumor measurements and molecular imaging methods, including FDG-PET. We have demonstrated that micro-PET analysis for conducting imaging studies in small animals allows for the noninvasive, quantitative, and repetitive imaging of biologic functions in living animals.20 In the current study, PET images were acquired sequentially using FDG to conduct a lesion-by-lesion comparison between control and experimental groups to determine treatment efficacy. Although the accuracy of detecting tumors >2 mm is high, animal PET analyses may miss approximately 35% of tumors <1 mm.20 Our animal PET scanner could not detect CCA tumors that measured <2 mm in size, and invasive CCA in the TAA rat model was not easily distinguished from the normal liver background because of the lack of distinct borders. Therefore, we used the average T/L ratio instead of the SUV of the tumor tissue alone as a quantitative method to measure tumor growth.20 We demonstrated a decrease in the mean T/L ratio of the SUVmean in NVP-AUY922-treated rats.

In conclusion, our bioinformatics analysis predicted that HSP90 inhibitors would be potential therapeutic agents in CCA. NVP-AUY922 had potent antitumor activity in the low-nanomolar range against CCA cells. In addition, NVP-AUY922 depleted the main client proteins involved in CCA oncogenesis and inhibited downstream proteins in the PI3K/AKT and KRAS/MAPK pathways. Tumor regression was demonstrated in the TAA-induced CCA animal model in rats that received NVP-AUY922. These preclinical data provide a rationale to conduct clinical trials with NVP-AUY922 in patients with CCA.

FUNDING SOURCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

This research was supported by grants from the Taiwan Cancer Clinic Foundation and the Yen Tjing Ling Medical Foundation to Taipei Veterans General Hospital (101DHA0100657 and 101DHA0100653 to M. Chen) and to Chang Gung Memorial Hospital (CMRPG390931 and CMRPG3B0361 to C. Yeh); by grants from the National Science Center (97-2314-B-182A-020-MY3 to C. Yeh and 100-2627-B-010-005-), the National Health Research Institutes (EX101-10029BI); National Taiwan Normal University (100NTNU-D-06), and the Ministry of Economic Affairs (100-EC-17-A-17-S1-152); and by a grant the Ministry of Education, Aim for the Top University Plan (National Yang Ming University) to C. Huang.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES