Detailed DNA methylation profiles of the E-cadherin promoter in the NCI-60 cancer cells

E-cadherin (E-cad) is a transmembrane adhesion glycoprotein, the expression of which is often reduced in invasive or metastatic tumors. To assess E-cad's distribution among different types of cancer cells, we used bisulfite-sequencing for detailed, base-by-base measurement of CpG methylation in E-cad's promoter region in the NCI-60 cell lines. The mean methylation levels of the cell lines were distributed bimodally, with values pushed toward either the high or low end of the methylation scale. The 38 epithelial cell lines showed substantially lower (28%) mean methylation levels compared with the nonepithelial cell lines (58%). The CpG site at -143 with respect to the transcriptional start was commonly methylated at intermediate levels, even in cell lines with low overall DNA methylation. We also profiled the NCI-60 cell lines using Affymetrix U133 microarrays and found E-cad expression to be correlated with E-cad methylation at highly statistically significant levels. Above a threshold of approximately 20% to 30% mean methylation, the expression of E-cad was effectively silenced. Overall, this study provides a type of detailed analysis of methylation that can also be applied to other cancer-related genes. As has been shown in recent years, DNA methylation status can serve as a biomarker for use in choosing therapy.

The LeFE algorithm: embracing the complexity of gene expression in the interpretation of microarray data

Interpretation of microarray data remains a challenge, and most methods fail to consider the complex, nonlinear regulation of gene expression. To address that limitation, we introduce Learner of Functional Enrichment (LeFE), a statistical/machine learning algorithm based on Random Forest, and demonstrate it on several diverse datasets: smoker/never smoker, breast cancer classification, and cancer drug sensitivity. We also compare it with previously published algorithms, including Gene Set Enrichment Analysis. LeFE regularly identifies statistically significant functional themes consistent with known biology.

Asparagine synthetase as a causal, predictive biomarker for L-asparaginase activity in ovarian cancer cells

L-Asparaginase (l-ASP), a bacterial enzyme used since the 1970s to treat acute lymphoblastic leukemia, selectively starves cells that cannot synthesize sufficient asparagine for their own needs. Molecular profiling of the NCI-60 cancer cell lines using five different microarray platforms showed strong negative correlations of asparagine synthetase (ASNS) expression and DNA copy number with sensitivity to l-ASP in the leukemia and ovarian cancer cell subsets. To assess whether the ovarian relationship is causal, we used RNA interference to silence ASNS in three ovarian lines and observed 4- to 5-fold potentiation of sensitivity to l-ASP with two of the lines. For OVCAR-8, the line that expresses the least ASNS, the potentiation was >500-fold. Significantly, that potentiation was >700-fold in the multidrug-resistant derivative OVCAR-8/ADR, showing that the causal relationship between ASNS expression and l-ASP activity survives development of classical multidrug resistance. Tissue microarrays confirmed low ASNS expression in a subset of clinical ovarian cancers as well as other tumor types. Overall, this pharmacogenomic/pharmacoproteomic study suggests the use of l-ASP for treatment of a subset of ovarian cancers (and perhaps other tumor types), with ASNS as a biomarker for patient selection.

Mutation analysis of 24 known cancer genes in the NCI-60 cell line set

The panel of 60 human cancer cell lines (the NCI-60) assembled by the National Cancer Institute for anticancer drug discovery is a widely used resource. The NCI-60 has been characterized pharmacologically and at the molecular level more extensively than any other set of cell lines. However, no systematic mutation analysis of genes causally implicated in oncogenesis has been reported. This study reports the sequence analysis of 24 known cancer genes in the NCI-60 and an assessment of 4 of the 24 genes for homozygous deletions. One hundred thirty-seven oncogenic mutations were identified in 14 (APC, BRAF, CDKN2, CTNNB1, HRAS, KRAS, NRAS, SMAD4, PIK3CA, PTEN, RB1, STK11, TP53, and VHL) of the 24 genes. All lines have at least one mutation among the cancer genes examined, with most lines (73%) having more than one. Identification of those cancer genes mutated in the NCI-60, in combination with pharmacologic and molecular profiles of the cells, will allow for more informed interpretation of anticancer agent screening and will enhance the use of the NCI-60 cell lines for molecularly targeted screens.

Depicting combinatorial complexity with the molecular interaction map notation

To help us understand how bioregulatory networks operate, we need a standard notation for diagrams analogous to electronic circuit diagrams. Such diagrams must surmount the difficulties posed by complex patterns of protein modifications and multiprotein complexes. To meet that challenge, we have designed the molecular interaction map (MIM) notation (http://discover.nci.nih.gov/mim/). Here we show the advantages of the MIM notation for three important types of diagrams: (1) explicit diagrams that define specific pathway models for computer simulation; (2) heuristic maps that organize the available information about molecular interactions and encompass the possible processes or pathways; and (3) diagrams of combinatorially complex models. We focus on signaling from the epidermal growth factor receptor family (EGFR, ErbB), a network that reflects the major challenges of representing in a compact manner the combinatorial complexity of multimolecular complexes. By comparing MIMs with other diagrams of this network that have recently been published, we show the utility of the MIM notation. These comparisons may help cell and systems biologists adopt a graphical language that is unambiguous and generally understood.

Integrating data on DNA copy number with gene expression levels and drug sensitivities in the NCI-60 cell line panel

Chromosome rearrangement, a hallmark of cancer, has profound effects on carcinogenesis and tumor phenotype. We used a panel of 60 human cancer cell lines (the NCI-60) as a model system to identify relationships among DNA copy number, mRNA expression level, and drug sensitivity. For each of 64 cancer-relevant genes, we calculated all 4,096 possible Pearson's correlation coefficients relating DNA copy number (assessed by comparative genomic hybridization using bacterial artificial chromosome microarrays) and mRNA expression level (determined using both cDNA and Affymetrix oligonucleotide microarrays). The analysis identified an association of ERBB2 overexpression with 3p copy number, a finding supported by data from human tumors and a mouse model of ERBB2-induced carcinogenesis. When we examined the correlation between DNA copy number for all 353 unique loci on the bacterial artificial chromosome microarray and drug sensitivity for 118 drugs with putatively known mechanisms of action, we found a striking negative correlation (-0.983; 95% bootstrap confidence interval, -0.999 to -0.899) between activity of the enzyme drug L-asparaginase and DNA copy number of genes near asparagine synthetase in the ovarian cancer cells. Previous analysis of drug sensitivity and mRNA expression had suggested an inverse relationship between mRNA levels of asparagine synthetase and L-asparaginase sensitivity in the NCI-60. The concordance of pharmacogenomic findings at the DNA and mRNA levels strongly suggests further study of L-asparaginase for possible treatment of a low-synthetase subset of clinical ovarian cancers. The DNA copy number database presented here will enable other investigators to explore DNA transcript-drug relationships in their own domains of research focus.

Chk2 molecular interaction map and rationale for Chk2 inhibitors

To organize the rapidly accumulating information on bioregulatory networks related to the histone gamma-H2AX-ATM-Chk2-p53-Mdm2 pathways in concise and unambiguous diagrams, we used the molecular interaction map notation (http://discover.nci.nih.gov/min). Molecular interaction maps are particularly useful for networks that include protein-protein binding and posttranslational modifications (e.g., phosphorylation). Both are important for nearly all of the proteins involved in DNA double-strand break signaling. Visualizing the regulatory circuits underlying cellular signaling may help identify key regulatory reactions and defects that can serve as targets for anticancer drugs.

Cancers as wounds that do not heal: differences and similarities between renal regeneration/repair and renal cell carcinoma

Cancers have been described as wounds that do not heal, suggesting that the two share common features. By comparing microarray data from a model of renal regeneration and repair (RRR) with reported gene expression in renal cell carcinoma (RCC), we asked whether those two processes do, in fact, share molecular features and regulatory mechanisms. The majority (77%) of the genes expressed in RRR and RCC were concordantly regulated, whereas only 23% were discordant (i.e., changed in opposite directions). The orchestrated processes of regeneration, involving cell proliferation and immune response, were reflected in the concordant genes. The discordant gene signature revealed processes (e.g., morphogenesis and glycolysis) and pathways (e.g., hypoxia-inducible factor and insulin-like growth factor-I) that reflect the intrinsic pathologic nature of RCC. This is the first study that compares gene expression patterns in RCC and RRR. It does so, in particular, with relation to the hypothesis that RCC resembles the wound healing processes seen in RRR. However, careful attention to the genes that are regulated in the discordant direction provides new insights into the critical differences between renal carcinogenesis and wound healing. The observations reported here provide a conceptual framework for further efforts to understand the biology and to develop more effective diagnostic biomarkers and therapeutic strategies for renal tumors and renal ischemia.

Selective toxicity of NSC73306 in MDR1-positive cells as a new strategy to circumvent multidrug resistance in cancer

ATP-binding cassette (ABC) proteins include the best known mediators of resistance to anticancer drugs. In particular, ABCB1 [MDR1/P-glycoprotein (P-gp)] extrudes many types of drugs from cancer cells, thereby conferring resistance to those agents. Attempts to overcome P-gp-mediated drug resistance using specific inhibitors of P-gp has had limited success and has faced many therapeutic challenges. As an alternative approach to using P-gp inhibitors, we characterize a thiosemicarbazone derivative (NSC73306) identified in a generic screen as a compound that exploits, rather than suppresses, P-gp function to induce cytotoxicity. Cytotoxic activity of NSC73306 was evaluated in vitro using human epidermoid, ovarian, and colon cancer cell lines expressing various levels of P-gp. Our findings suggest that cells become hypersensitive to NSC73306 in proportion to the increased P-gp function and multidrug resistance (MDR). Abrogation of both sensitivity to NSC73306 and resistance to P-gp substrate anticancer agents occurred with specific inhibition of P-gp function using either a P-gp inhibitor (PSC833, XR9576) or RNA interference, suggesting that cytotoxicity was linked to MDR1 function, not to other, nonspecific factors arising during the generation of resistant or transfected cells. Molecular characterization of cells selected for resistance to NSC73306 revealed loss of P-gp expression and consequent loss of the MDR phenotype. Although hypersensitivity to NSC73306 required functional expression of P-gp, biochemical assays revealed no direct interaction between NSC73306 and P-gp. This article shows that NSC73306 kills cells with intrinsic or acquired P-gp-induced MDR and indirectly acts to eliminate resistance to MDR1 substrates.

Sequencing and analysis of 10,967 full-length cDNA clones from Xenopus laevis and Xenopus tropicalis reveals post-tetraploidization transcriptome remodeling

Sequencing of full-insert clones from full-length cDNA libraries from both Xenopus laevis and Xenopus tropicalis has been ongoing as part of the Xenopus Gene Collection Initiative. Here we present 10,967 full ORF verified cDNA clones (8049 from X. laevis and 2918 from X. tropicalis) as a community resource. Because the genome of X. laevis, but not X. tropicalis, has undergone allotetraploidization, comparison of coding sequences from these two clawed (pipid) frogs provides a unique angle for exploring the molecular evolution of duplicate genes. Within our clone set, we have identified 445 gene trios, each comprised of an allotetraploidization-derived X. laevis gene pair and their shared X. tropicalis ortholog. Pairwise dN/dS, comparisons within trios show strong evidence for purifying selection acting on all three members. However, dN/dS ratios between X. laevis gene pairs are elevated relative to their X. tropicalis ortholog. This difference is highly significant and indicates an overall relaxation of selective pressures on duplicated gene pairs. We have found that the paralogs that have been lost since the tetraploidization event are enriched for several molecular functions, but have found no such enrichment in the extant paralogs. Approximately 14% of the paralogous pairs analyzed here also show differential expression indicative of subfunctionalization.

AbMiner: a bioinformatic resource on available monoclonal antibodies and corresponding gene identifiers for genomic, proteomic, and immunologic studies

Background

Monoclonal antibodies are used extensively throughout the biomedical sciences for detection of antigens, either in vitro or in vivo. We, for example, have used them for quantitation of proteins on "reverse-phase" protein lysate arrays. For those studies, we quality-controlled > 600 available monoclonal antibodies and also needed to develop precise information on the genes that encode their antigens. Translation among the various protein and gene identifier types proved non-trivial because of one-to-many and many-to-one relationships. To organize the antibody, protein, and gene information, we initially developed a relational database in Filemaker for our own use. When it became apparent that the information would be useful to many other researchers faced with the need to choose or characterize antibodies, we developed it further as AbMiner, a fully relational web-based database under MySQL, programmed in Java.

Description

AbMiner is a user-friendly, web-based relational database of information on > 600 commercially available antibodies that we validated by Western blot for protein microarray studies. It includes many types of information on the antibody, the immunogen, the vendor, the antigen, and the antigen's gene. Multiple gene and protein identifier types provide links to corresponding entries in a variety of other public databases, including resources for phosphorylation-specific antibodies. AbMiner also includes our quality-control data against a pool of 60 diverse cancer cell types (the NCI-60) and also protein expression levels for the NCI-60 cells measured using our high-density "reverse-phase" protein lysate microarrays for a selection of the listed antibodies. Some other available database resources give information on antibody specificity for one or a couple of cell types. In contrast, the data in AbMiner indicate specificity with respect to the antigens in a pool of 60 diverse cell types from nine different tissues of origin.

Conclusion

AbMiner is a relational database that provides extensive information from our own laboratory and other sources on more than 600 available antibodies and the genes that encode the antibodies' antigens. The data will be made freely available at

Comparison of Methods for Sequential Screening of Large Compound Sets

Sequential screening is an iterative procedure that can greatly increase hit rates over random screening or non-iterative procedures. We studied the effects of three factors on enrichment rates: the method used to rank compounds, the molecular descriptor set and the selection of initial training set. The primary factor influencing recovery rates was the method of selecting the initial training set. Rates for recovering active compounds were substantially lower with the diverse training sets than they were with training sets selected by other methods. Because structure-activity information is incrementally enhanced in intermediate training sets, sequential screening provides significant improvement in the average rate of recovery of active compounds when compared with non-iterative selection procedures.

The p53 tumor suppressor network is a key responder to microenvironmental components of chronic inflammatory stress

Activation of the p53 network plays a central role in the inflammatory stress response associated with ulcerative colitis and may modulate cancer risk in patients afflicted with this chronic disease. Here, we describe the gene expression profiles associated with four microenvironmental components of the inflammatory response (NO*, H2O2, DNA replication arrest, and hypoxia) that result in p53 stabilization and activation. Isogenic HCT116 and HCT116 TP53-/- colon cancer cells were exposed to the NO* donor Sper/NO, H2O2, hypoxia, or hydroxyurea, and their mRNA was analyzed using oligonucleotide microarrays. Overall, 1,396 genes changed in a p53-dependent manner (P < 0.001), with the majority representing a "unique" profile for each condition. Only 14 genes were common to all four conditions. Included were eight known p53 target genes. Hierarchical sample clustering distinguished early (1 and 4 hours) from late responses (8, 12, and 24 hours), and each treatment was differentiated from the others. Overall, NO* and hypoxia stimulated similar transcriptional responses. Gene ontology analysis revealed cell cycle as a key feature of stress responses and confirmed the similarity between NO* and hypoxia. Cell cycle profiles analyzed by flow cytometry showed that NO* and hypoxia induced quiescent S-phase and G2-M arrest. Using a novel bioinformatic algorithm, we identified several putative p53-responsive elements among the genes induced in a p53-dependent manner, including four [KIAA0247, FLJ12484, p53CSV (HSPC132), and CNK (PLK3)] common to all exposures. In summary, the inflammatory stress response is a complex, integrated biological network in which p53 is a key molecular node regulating gene expression.

Molecular interaction maps of bioregulatory networks: a general rubric for systems biology

A standard for bioregulatory network diagrams is urgently needed in the same way that circuit diagrams are needed in electronics. Several graphical notations have been proposed, but none has become standard. We have prepared many detailed bioregulatory network diagrams using the molecular interaction map (MIM) notation, and we now feel confident that it is suitable as a standard. Here, we describe the MIM notation formally and discuss its merits relative to alternative proposals. We show by simple examples how to denote all of the molecular interactions commonly found in bioregulatory networks. There are two forms of MIM diagrams. "Heuristic" MIMs present the repertoire of interactions possible for molecules that are colocalized in time and place. "Explicit" MIMs define particular models (derived from heuristic MIMs) for computer simulation. We show also how pathways or processes can be highlighted on a canonical heuristic MIM. Drawing a MIM diagram, adhering to the rules of notation, imposes a logical discipline that sharpens one's understanding of the structure and function of a network.

Biomarkers in cancer staging, prognosis and treatment selection

Advances in genomics, proteomics and molecular pathology have generated many candidate biomarkers with potential clinical value. Their use for cancer staging and personalization of therapy at the time of diagnosis could improve patient care. However, translation from bench to bedside outside of the research setting has proved more difficult than might have been expected. Understanding how and when biomarkers can be integrated into clinical care is crucial if we want to translate the promise into reality.

Nova regulates brain-specific splicing to shape the synapse

Alternative RNA splicing greatly increases proteome diversity and may thereby contribute to tissue-specific functions. We carried out genome-wide quantitative analysis of alternative splicing using a custom Affymetrix microarray to assess the role of the neuronal splicing factor Nova in the brain. We used a stringent algorithm to identify 591 exons that were differentially spliced in the brain relative to immune tissues, and 6.6% of these showed major splicing defects in the neocortex of Nova2-/- mice. We tested 49 exons with the largest predicted Nova-dependent splicing changes and validated all 49 by RT-PCR. We analyzed the encoded proteins and found that all those with defined brain functions acted in the synapse (34 of 40, including neurotransmitter receptors, cation channels, adhesion and scaffold proteins) or in axon guidance (8 of 40). Moreover, of the 35 proteins with known interaction partners, 74% (26) interact with each other. Validating a large set of Nova RNA targets has led us to identify a multi-tiered network in which Nova regulates the exon content of RNAs encoding proteins that interact in the synapse.

High-Throughput GoMiner, an 'industrial-strength' integrative gene ontology tool for interpretation of multiple-microarray experiments, with application to studies of Common Variable Immune Deficiency (CVID)

BACKGROUND

We previously developed GoMiner, an application that organizes lists of 'interesting' genes (for example, under-and overexpressed genes from a microarray experiment) for biological interpretation in the context of the Gene Ontology. The original version of GoMiner was oriented toward visualization and interpretation of the results from a single microarray (or other high-throughput experimental platform), using a graphical user interface. Although that version can be used to examine the results from a number of microarrays one at a time, that is a rather tedious task, and original GoMiner includes no apparatus for obtaining a global picture of results from an experiment that consists of multiple microarrays. We wanted to provide a computational resource that automates the analysis of multiple microarrays and then integrates the results across all of them in useful exportable output files and visualizations.

RESULTS

We now introduce a new tool, High-Throughput GoMiner, that has those capabilities and a number of others: It (i) efficiently performs the computationally-intensive task of automated batch processing of an arbitrary number of microarrays, (ii) produces a human-or computer-readable report that rank-orders the multiple microarray results according to the number of significant GO categories, (iii) integrates the multiple microarray results by providing organized, global clustered image map visualizations of the relationships of significant GO categories, (iv) provides a fast form of 'false discovery rate' multiple comparisons calculation, and (v) provides annotations and visualizations for relating transcription factor binding sites to genes and GO categories.

CONCLUSION

High-Throughput GoMiner achieves the desired goal of providing a computational resource that automates the analysis of multiple microarrays and integrates results across all of the microarrays. For illustration, we show an application of this new tool to the interpretation of altered gene expression patterns in Common Variable Immune Deficiency (CVID). High-Throughput GoMiner will be useful in a wide range of applications, including the study of time-courses, evaluation of multiple drug treatments, comparison of multiple gene knock-outs or knock-downs, and screening of large numbers of chemical derivatives generated from a promising lead compound.

Quality assessment of microarrays: visualization of spatial artifacts and quantitation of regional biases

BACKGROUND

Quality-control is an important issue in the analysis of gene expression microarrays. One type of problem is regional bias, in which one region of a chip shows artifactually high or low intensities (or ratios in a two-channel array) relative to the majority of the chip. Current practice in quality assessment for microarrays does not address regional biases.

RESULTS

We present methods implemented in R for visualizing regional biases and other spatial artifacts on spotted microarrays and Affymetrix chips. We also propose a statistical index to quantify regional bias and investigate its typical distribution on spotted and Affymetrix arrays. We demonstrate that notable regional biases occur on both Affymetrix and spotted arrays and that they can make a significant difference in the case of spotted microarray results. Although strong biases are also seen at the level of individual probes on Affymetrix chips, the gene expression measures are less affected, especially when the RMA method is used to summarize intensities for the probe sets. A web application program for visualization and quantitation of regional bias is provided at http://www.discover.nci.nih.gov/affytools.

CONCLUSION

Researchers should visualize and measure the regional biases and should estimate their impact on gene expression measurements obtained. Here, we (i) introduce pictorial visualizations of the spatial biases; (ii) present for Affymetrix chips a useful resolution of the biases into two components, one related to background, the other to intensity scale factor; (iii) introduce a single parameter to reflect the global bias present across an array. We also examine the pattern distribution of such biases and conclude that algorithms based on smoothing are unlikely to compensate adequately for them.

Integromic analysis of the NCI-60 cancer cell lines

Microarray-based transcript profiling has become exceedingly popular, particularly for breast cancer. However, other 'omic' profiling technologies at the DNA, RNA, protein, functional, and pharmacological levels are also becoming increasingly practical. We define 'integromics' as the melding of such diverse types of data from different experimental platforms. The whole can sometimes be more than the sum of its parts. We describe here a set of integromic studies in which we have profiled the 60 human cancer cell lines (the NCI-60) used by the National Cancer Institute to screen >100,000 chemical compounds over the last 13 years. Patterns of potency in the screen can be mapped into molecular structures of the compounds or into molecular characteristics of the cells. Here we discuss conceptual and experimental aspects of the profiling, as well as a number of bioinformatic computer programs (CIMminer, MedMiner, MatchMiner, and GoMiner) that we have developed for biological interpretation of the profiles. As briefly reviewed here, we have used the combination of NCI-60 data types to identify markers for distinguishing tumor types and to obtain pharmacogenomic clues for possible individualization of a cancer therapy.

Workshop on Cancer Biometrics: Identifying Biomarkers and Surrogates of Cancer in Patients

The current excitement about molecular targeted therapies has driven much of the recent dialog in cancer diagnosis and treatment. Particularly in the biologic therapy of cancer, identifiable antigenic T-cell targets restricted by MHC molecules and the related novel stress molecules such as MICA/B and Letal allow a degree of precision previously unknown in cancer therapy. We have previously held workshops on immunologic monitoring and angiogenesis monitoring. This workshop was designed to discuss the state of the art in identification of biomarkers and surrogates of tumor in patients with cancer, with particular emphasis on assays within the blood and tumor. We distinguish this from immunologic monitoring in the sense that it is primarily a measure of the tumor burden as opposed to the immune response to it. Recommendations for intensive investigation and targeted funding to enable such strategies were developed in seven areas: genomic analysis; detection of molecular markers in peripheral blood and lymph node by tumor capture and RT-PCR; serum, plasma, and tumor proteomics; immune polymorphisms; high content screening using flow and imaging cytometry; immunohistochemistry and tissue microarrays; and assessment of immune infiltrate and necrosis in tumors. Concrete recommendations for current application and enabling further development in cancer biometrics are summarized. This will allow a more informed, rapid, and accurate assessment of novel cancer therapies.

Karyotypic "state" as a potential determinant for anticancer drug discovery

Cancer is a genetic disease caused by genomic instability. In many cancers, this instability is manifested by chromosomal reconfigurations and karyotypic complexity. These features are particular hallmarks of the epithelial cancers that are some of the malignancies most resistant to long term control by current chemotherapeutic agents. We have asked whether we could use karyotypic complexity and instability as determinants for the screening of potential anticancer compounds. Using a panel of well characterized cancer cell lines, we have been able to identify specific groups of chemical compounds that are more cytotoxic toward the relatively more karyotypically complex and unstable panel members. Thus, we delineate an approach for the identification of "lead compounds" for anticancer drug discovery complementary to those that are focused at the outset on a given gene or pathway.

Analysis of ATP-Binding Cassette Transporter Expression in Drug-Selected Cell Lines by a Microarray Dedicated to Multidrug Resistance

Discovery of the multidrug resistance protein 1 (MDR1), an ATP-binding cassette (ABC) transporter able to transport many anticancer drugs, was a clinically relevant breakthrough in multidrug resistance research. Although the overexpression of ABC transporters such as P-glycoprotein/ABCB1, MRP1/ABCC1, and MXR/ABCG2 seems to be a major cause of failure in the treatment of cancer, acquired resistance to multiple anticancer drugs may also be multifactorial, involving alteration of detoxification processes, apoptosis, DNA repair, drug uptake, and overexpression of other ABC transporters. As a tool for the study of such phenomena, we designed and created a microarray platform, the ABC-ToxChip, to evaluate relative levels of transcriptional activation among genes involved in the various mechanisms of resistance. In the ABC-ToxChip, a comprehensive set of genes important in toxicological responses (represented by 2200 cDNA probes) is complemented with probes specifically matching ABC transporters as well as oligonucleotides representing 18,000 unique human genes. By comparing the transcriptional profiles of KB-3-1 and DU-145 parental cells with resistant derivatives selected in colchicine (KB-8-5), and 9-nitro-camptothecin (RCO.1), respectively, we demonstrate that ABC transporters (ABCB1/MDR1 and ABCC2/MRP2, respectively) show dramatic overexpression, whereas the glutathione S-transferase gene GST-Pi shows the strongest decrease in expression among the 20,000 genes studied. The results were confirmed by quantitative reverse transcription-polymerase chain reaction and immunohistochemistry. The custom-designed ABC-Tox microarray presented here will be helpful to elucidate mechanisms leading to anticancer drug resistance.