ORIGINAL ARTICLE Gene Expression Signatures Associated with The
Total Page:16
File Type:pdf, Size:1020Kb
Leukemia (2006) 20, 1542–1550 & 2006 Nature Publishing Group All rights reserved 0887-6924/06 $30.00 www.nature.com/leu ORIGINAL ARTICLE Gene expression signatures associated with the resistance to imatinib Y-J Chung1,6, T-M Kim1,6, D-W Kim2, H Namkoong3, HK Kim3, S-A Ha3, S Kim3, SM Shin3, J-H Kim4, Y-J Lee4, H-M Kang1 and JW Kim3,5 1Department of Microbiology, College of Medicine, The Catholic University of Korea, Seoul, Korea; 2Department of Internal medicine, College of Medicine, The Catholic University of Korea, Seoul, Korea; 3Molecular Genetic Laboratory, College of Medicine, The Catholic University of Korea, Seoul, Korea; 4Seoul National University Biomedical Informatics (SNUBI) and Human Genome Research Institute, College of Medicine, Seoul National University, Seoul, Korea; and 5Department of Obstetrics and Gynecology, College of Medicine, The Catholic University of Korea, Seoul, Korea Imatinib (imatinib mesylate, STI-571, Gleevec) is a selective Transcriptional amplification or point mutations in the BCR-ABL tyrosine kinase inhibitor that has been used as a tyrosine kinase domain of BCR-ABL have been frequently highly effective chemoagent for treating chronic myelogenous 3–5 leukemia. However, the initial response to imatinib is often observed in resistant clinical cases or in vitro cell models. followed by the recurrence of a resistant form of the disease, Although these changes are considered almost sufficient cause which is major obstacle to many therapeutic modalities. The of obtaining imatinib-resistance, previous studies have also aim of this study was to identify the gene expression signatures shown that resistance can occur without any apparent ampli- that confer resistance to imatinib. A series of four resistant fication or point mutations in BCR-ABL.6–8 In addition, the K562 sublines was established with different imatinib dosage overexpression of multidrug resistance genes has been suggested (200, 400, 600 and 800 nM) and analyzed using microarray 9,10 technology. The transcripts of the genes showing universal or to confer the resistance to imatinib. These reports suggest dose-dependent expression changes across the resistant that heterogeneous mechanisms might be responsible for sublines were identified. The gene sets associated with the imatinib-resistance. To identify key genetic elements responsi- imatinib-resistance were also identified using gene set enrich- ble for imatinib-resistance, various approaches including global ment analysis. In the resistant K562 sublines, the transcription- gene expression analysis using a microarray have been and apoptosis-related expression signatures were upregulated, used.7,11,12 whereas those related to the protein and energy metabolism were downregulated. Several genes identified in this study With the aim of deciphering gene expression signatures such as IGF1 and RAB11A have the potential to become associated with imatinib-resistance, this study focused on the surrogate markers useful in a clinical evaluation of imatinib- transcripts showing biologically relevant expression changes by resistant patients without BCR-ABL mutation. The expression examining the expression profiles across four imatinib-resistant signatures identified in this study provide insights into the sublines established with different imatinib dosage and primary mechanism of imatinib-resistance and are expected to facilitate CML cases. We present expression profiles and candidate signal the development of an effective diagnostic and therapeutic strategy. pathways associated with imatinib-resistance through improved Leukemia (2006) 20, 1542–1550. doi:10.1038/sj.leu.2404310; bioinformatical approaches including gene set enrichment published online 20 July 2006 analysis. Keywords: drug resistance; expression array; imatinib mesylate; STI-571 Materials and methods Establishment of imatinib-resistant cell lines The resistance to imatinib was established by exposing the erythroid leukemic K562 cell lines to increasing imatinib Introduction concentrations by 50 nM every 2 weeks. During the culture with imatinib, a series of four resistant sublines exposed to Imatinib (imatinib mesylate, STI-571, Gleevec) is a selective various imatinib concentrations (200, 400, 600 and 800 nM) tyrosine kinase inhibitor that has been successfully used to treat were established. Any mutations of the BCR-ABL kinase domain chronic myloid leukemia (CML). Targeting to BCR-ABL fusion were screened as described previously.4 In addition, the kinase, the primary pathogenic molecule for disease, imatinib expression level of BCR-ABL kinase was examined using real- achieved remarkable remission rates up to B80%, especially for 1 time reverse transcriptase-polymerase chain reaction (PCR) with initial phase of CML. However, relapse after the initial the same primers used for mutation analysis and for human hematologic and cytologenetic response frequently occurred 2 glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as inter- in late-stage disease. This suggests that BCR-ABL targeted nal control. therapy can be hindered by the acquisition of resistance, which seriously reduces the clinical efficacy of imatinib. RNA preparation and hybridization Correspondence: Professor JW Kim, Molecular Genetic Laboratory, Total RNA of K562 and imatinib-resistant sublines was isolated College of Medicine, The Catholic University of Korea, Seoul 137- with Trizol (Invitrogen, Carlsbad, CA) according to the 040, Korea. manufacturer’s instruction. The quantity and quality of extracted E-mail: [email protected] 6These authors contributed equally to this work RNA was assessed by using a Bioanalyzer 2100 (Agilent Received 20 January 2006; revised 23 May 2006; accepted 6 June Technologies, Waldbronn, Germany). Applied Biosystems Human 2006; published online 20 July 2006 Genome Survey Microarray Version 2.0 representing a set Expression signatures associated with imatinib-resistant phenotype Y-J Chung et al 1543 of B30 000 human genes were used to analyze the expression rate methods.17 The transcripts significant (qo0.05) for up- and profiles of the five cell lines (parental sensitive K562 and four downprogression were further tested for the significance in 600 resistant sublines). Digoxigenin-UTP-labeled cRNA was gener- and 800 nM t-tests with Welch correction and examined to be ated and linearly amplified from 5 mg of total RNA using Applied coordinately changed across resistant sublines. Detailed de- Biosystems Chemiluminescent RT-IVT Labeling Kit V2.0. Array scription with used parameters is available in Supplementary hybridization, Chemiluminescence detection, image acquisition note 1. and analysis were performed using Applied Biosystems Chemi- luminescence Detection Kit and Applied Biosystems 1700 Chemiluminescent Microarray Analyzer according to the Gene set enrichment analysis manufacturer’s protocol. The hybridization reaction was tripli- Functionally related genes were retrieved from public gene cated per sample. database: Gene Ontology, BioCarta (http://www.biocarta.com), KEGG (Kyoto Encyclopedia of Genes and Genomes) and GenMAPP (Gene MicroArray Pathway Profiler).18–20 According Data processing to gene-versus-annotation matches, genes were categorized into Image of each array hybridization was collected using the 1700 annotation-specified gene sets. A total of 3774 gene sets were analyzer, which is equipped with high-resolution, large-format collected, whose members have a common functional annota- CCD camera, including two ‘short’ chemiluminescent images tion and are not necessarily exclusive. Enrichment analysis was with 5 s exposure length for gene expression analysis, two performed using parametrical analysis of gene enrichment that fluorescent images for feature finding and spot normalization directly calculates the significance of enrichment based on and 2QC images for spectrum cross-talk correction. Images Z-statistics.21 To perform enrichment analysis, we further were auto-gridded and the chemiluminescent signals were restricted initial gene sets into 721 sets containing at least 10 quantified. After the background subtraction, spot intensity data genes in minimal gene set.21 We used two kinds of parametrical were normalized by using variance stabilization and normal- 13 values, relative difference r.d. and Fk to calculate the signifi- ization method. cance level based on Z-statistics. Total six enrichment analyses were performed using four sets of r.d. representing comparisons between four resistant sublines and control K562 and two sets of Determination of differentially expressed transcripts One-way analysis of variance analysis was performed on the Fk representing up- and downprogression. We determined average expression of 29 098 genes across five cell lines activated gene sets when they are significantly enriched in top including those of parental K562 cell lines. The most variable r.d. for both comparisons of 600 and 800 nM and also in top Fk in 3000 transcripts were chosen and hierarchical clustering via upprogression and vice versa for repressed gene sets. average linking was performed using custom software, Cluster and Treeview (http://rana.lbl.gov/EisenSoftware.htm).14 To de- termine whether a transcript is up- or downregulated in resistant Real-time quantitative PCR assay sublines, the expression change compared to sensitive parental Total RNA from blood samples of six CML patients before and control was measured by calculating the relative difference, r.d. after acquiring imatinib-resistance (kindly donated from Korean as following: Leukemia