Fuzzy-logic based strategy for validation of multiplex methods: example with qualitative GMO assays.

This paper illustrates the advantages that a fuzzy-based aggregation method could bring into the validation of a multiplex method for GMO detection (DualChip GMO kit, Eppendorf). Guidelines for validation of chemical, bio-chemical, pharmaceutical and genetic methods have been developed and ad hoc validation statistics are available and routinely used, for in-house and inter-laboratory testing, and decision-making. Fuzzy logic allows summarising the information obtained by independent validation statistics into one synthetic indicator of overall method performance. The microarray technology, introduced for simultaneous identification of multiple GMOs, poses specific validation issues (patterns of performance for a variety of GMOs at different concentrations). A fuzzy-based indicator for overall evaluation is illustrated in this paper, and applied to validation data for different genetically modified elements. Remarks were drawn on the analytical results. The fuzzy-logic based rules were shown to be applicable to improve interpretation of results and facilitate overall evaluation of the multiplex method.

The MicroArray Quality Control (MAQC) project shows inter- and intraplatform reproducibility of gene expression measurements.

Over the last decade, the introduction of microarray technology has had a profound impact on gene expression research. The publication of studies with dissimilar or altogether contradictory results, obtained using different microarray platforms to analyze identical RNA samples, has raised concerns about the reliability of this technology. The MicroArray Quality Control (MAQC) project was initiated to address these concerns, as well as other performance and data analysis issues. Expression data on four titration pools from two distinct reference RNA samples were generated at multiple test sites using a variety of microarray-based and alternative technology platforms. Here we describe the experimental design and probe mapping efforts behind the MAQC project. We show intraplatform consistency across test sites as well as a high level of interplatform concordance in terms of genes identified as differentially expressed. This study provides a resource that represents an important first step toward establishing a framework for the use of microarrays in clinical and regulatory settings.

The balance of reproducibility, sensitivity, and specificity of lists of differentially expressed genes in microarray studies.

BACKGROUND: Reproducibility is a fundamental requirement in scientific experiments. Some recent publications have claimed that microarrays are unreliable because lists of differentially expressed genes (DEGs) are not reproducible in similar experiments. Meanwhile, new statistical methods for identifying DEGs continue to appear in the scientific literature. The resultant variety of existing and emerging methods exacerbates confusion and continuing debate in the microarray community on the appropriate choice of methods for identifying reliable DEG lists. RESULTS: Using the data sets generated by the MicroArray Quality Control (MAQC) project, we investigated the impact on the reproducibility of DEG lists of a few widely used gene selection procedures. We present comprehensive results from inter-site comparisons using the same microarray platform, cross-platform comparisons using multiple microarray platforms, and comparisons between microarray results and those from TaqMan - the widely regarded "standard" gene expression platform. Our results demonstrate that (1) previously reported discordance between DEG lists could simply result from ranking and selecting DEGs solely by statistical significance (P) derived from widely used simple t-tests; (2) when fold change (FC) is used as the ranking criterion with a non-stringent P-value cutoff filtering, the DEG lists become much more reproducible, especially when fewer genes are selected as differentially expressed, as is the case in most microarray studies; and (3) the instability of short DEG lists solely based on P-value ranking is an expected mathematical consequence of the high variability of the t-values; the more stringent the P-value threshold, the less reproducible the DEG list is. These observations are also consistent with results from extensive simulation calculations. CONCLUSION: We recommend the use of FC-ranking plus a non-stringent P cutoff as a straightforward and baseline practice in order to generate more reproducible DEG lists. Specifically, the P-value cutoff should not be stringent (too small) and FC should be as large as possible. Our results provide practical guidance to choose the appropriate FC and P-value cutoffs when selecting a given number of DEGs. The FC criterion enhances reproducibility, whereas the P criterion balances sensitivity and specificity.

ABCA3 as a possible cause of drug resistance in childhood acute myeloid leukemia.

BACKGROUND: A major issue in the treatment of acute myeloid leukemia (AML) is resistance to chemotherapeutic drugs. Multidrug resistance can be caused by ATP-binding cassette (ABC) transporters that function as drug efflux pumps. The majority of these proteins have not yet been examined in malignant diseases. EXPERIMENTAL DESIGN: A newly developed microarray for the simultaneous quantification of 38 ABC transporter genes and Taqman real-time PCR was used to analyze the expression of ABC transporters in pediatric AML and healthy bone marrow. Small interfering RNA was used to verify the role of ABCA3 in drug resistance. RESULTS: Using the microarray, we identified four new ABC transporters, which were overexpressed in many AML samples compared with healthy bone marrow: ABCA2, ABCA3, ABCB2, and ABCC10. The overexpression of these four genes was verified by real-time PCR in 42 samples from children with AML and 18 samples of healthy bone marrow. The median expression of ABCA3 was three times higher in 21 patients who had failed to achieve remission after the first course of chemotherapy than in a well-matched group of 21 patients who had achieved remission at this stage (P = 0.023). Incubation of cell lines with a number of different cytostatic drugs induced an up-regulation of ABCA3. Down-regulation of ABCA3 by small interfering RNA sensitized cells to doxorubicin. CONCLUSION: Our results show that ABCA2, ABCA3, ABCB2, and ABCC10 are overexpressed in childhood AML compared with healthy bone marrow. ABCA3 is the most likely transporter to cause drug resistance.

Expression profiling of ATP-binding cassette transporters in childhood T-cell acute lymphoblastic leukemia.

A major issue in the treatment of T-cell acute lymphoblastic leukemia (T-ALL) is resistance to chemotherapeutic drugs. Multidrug resistance can be caused by ATP-binding cassette (ABC) transporters. The majority of these proteins have not yet been examined in T-ALL. Using a newly developed microarray for the simultaneous quantification of 38 ABC transporter genes, we observed a consistent overexpression of ABCA2/ABCA3 in clinical samples of ALL. Therefore, we analyzed the association of these two genes with drug resistance. Treatment of CCRF-CEM and Jurkat cells with methotrexate, vinblastine, or doxorubicin led to an induction of ABCA3 expression, whereas a significant increase of ABCA2 expression was only observed in Jurkat cells. To study the causal relationship of ABCA2/A3 overexpression with drug resistance, we applied RNA interference (RNAi) technology. RNAi specific for ABCA2 or ABCA3 led to a partial decrease of expression in these two ABC transporters. Upon cotreatment of RNAi for ABCA2 with methotrexate and vinblastine, a partial decrease of ABCA2 expression as well as a simultaneous increase of ABCA3 expression was observed. Vice versa, ABCA3 RNAi plus drugs decreased ABCA3 and increased ABCA2 expression. This indicates that down-regulation of one ABC transporter was compensated by the up-regulation of the other. Application of RNAi for both ABCA2 and ABCA3 resulted in a more efficient reduction of the expression of both transporters. As a consequence, a significant sensitization of cells to cytostatic drugs was achieved. In conclusion, ABCA2 and ABCA3 are expressed in many T-ALL and contribute to drug resistance.

Microarray-based detection of multidrug resistance in human tumor cells by expression profiling of ATP-binding cassette transporter genes.

Different mechanisms of drug resistance, including ATP-binding cassette (ABC) transporters, are responsible for treatment failure of tumors. We developed a low-density DNA microarray which contains 38 genes of the ABC transporter gene family. This tool has been validated with three different multidrug-resistant sublines (CEM/ADR5000, HL60/AR, and MCF7/CH1000) known to overexpress either the ABCB1 (MDR1), ABCC1 (MRP1), or ABCG2 (MXR and BCRP) genes. When compared with their drug-sensitive parental lines, we observed not only the overexpression of these genes in the multidrug-resistant cell lines but also of other ABC transporter genes pointing to their possible role in multidrug resistance. These results were corroborated by quantitative real-time reverse transcription-PCR. As the microarray allows the determination of the expression profile of many ABC transporters in a single hybridization experiment, it may be useful as a diagnostic tool to detect drug resistance in clinical samples.

Comparison of replicative senescence and stress-induced premature senescence combining differential display and low-density DNA arrays.

Human diploid fibroblasts (HDFs) exposed to subcytotoxic stress display many features of senescence. Using differential display RT-PCR, gene expression of HDFs in premature senescence induced by tert-butylhydroperoxide or ethanol and in replicative senescence was compared to gene expression of HDFs at early cumulative population doublings. Thirty genes of known function were identified from the 265 differentially displayed cDNA fragments. A customized low-density array allowed to confirm the relative level of the corresponding 30 transcripts. We found differential expression of genes coding for proteins implicated namely in growth arrest (PTEN, IGFBP-3, LRP-1 and CAV1), senescent morphogenesis (TGF-beta1 and LOXL2) and iron metabolism (TFR and FTL).

Molecular characterization of breast cancer cell lines by a low-density microarray.

We designed a low-density microarray carrying 132 DNA capture sequences highly specific for genes known to be differentially expressed among breast tumors and BCC lines or associated with specific tumor properties (cell-cycle alteration, proteolysis, adhesion, hormone sensitivity, etc). We analyzed gene expression in 11 BCC lines among which 6 had already been extensively studied (BT-474, Hs578T, MCF-7, MDA-MB-231, MDA-MB-453, T-47D) and 5 were still poorly characterized (Evsa-T, IBEP-1, IBEP-2, IBEP-3, KPL-1). Some data obtained were verified or extended by real-time polymerase chain reaction (real-time PCR), Northern blotting, Western blotting, immunohistochemistry and cell growth studies. Clustering analysis of the low-density microarray data allowed the sorting of BCC lines into two classes and supported a major discriminatory role for ER alpha, confirming data from previous studies. A few genes that are highly and specifically expressed in one cell line were identified, such as MGB1 (mammaglobin 1) in Evsa-T cells, and PIP (prolactin-inducible protein) in MDA-MB-453 BCC, suggesting an apocrine origin for these latter cells. Two BCC lines (IBEP-1 and IBEP-3) that had been previously characterized as ER alpha-negative, were classified by the low-density microarray among ER alpha-positive lines (MCF-7, T-47D, IBEP-2, BT-474, KPL-1) and were indeed confirmed as receptor-positive (at both mRNA and protein levels) and hormone-responsive cells. In conclusion, our results support the utility of a low-density microarray approach in cases where the cost and exhaustiveness of high-density microarrays may constitute a drawback; for instance, in obtaining a rapid phenotype evaluation in cell populations freshly isolated from breast tumors.

Gene expression profiling of drug metabolism and toxicology markers using a low-density DNA microarray.

DNA microarrays are useful tools to study changes of gene expression in response to a treatment with drugs. Here, we describe the optimization of conditions for the cDNA synthesis and hybridization protocols to be used for a low-density DNA microarray called 'Rat HepatoChips.' This DNA microarray with 59 carefully selected genes could be used to study changes in gene expression levels due to a treatment with xenobiotic. These 59 genes (including 8 housekeeping genes) have been selected among potential toxic markers involved in basic cellular processes and drug metabolism related genes. Using the optimized conditions, the results were shown to be reproducible, with 6% variation between the duplicated spots and 10% between arrays. Conditions were optimized to allow quantification with a dynamic range of four log units. In order to demonstrate the major advantage of these tool for studying gene expression, samples of control rat liver were compared with those of animals dosed with phenobarbital (PB) or pregnenolone-16 alpha-carbonitrile (PCN), two compounds well known to induce cytochrome P450 isoforms of 2B and 3A subfamilies, respectively. This microarray has shown that other genes apart from the corresponding CYP P450 genes have been changed due to PB and PCN treatment. Apoptosis-related genes have shown to be changed due to PB and PCN treatment, which confirms results from previous work.

Use of a low-density microarray for studying gene expression patterns induced by hepatotoxicants on primary cultures of rat hepatocytes.

In the field of gene expression analysis, DNA microarray technology is having a major impact on many different areas including toxicology. For instance, a number of studies have shown that transcription profiling can generate the information needed to assign a compound to a mode-of-action class. In this study, we investigated whether compounds inducing similar toxicological endpoints produce similar changes in gene expression. In vitro primary rat hepatocytes were exposed to 11 different hepatotoxicants: acetaminophen, amiodarone, clofibrate, erythromycin estolate, isoniazid, alpha-naphtylylisothiocyanate, beta-naphtoflavone, 4-pentenoic acid, phenobarbital, tetracycline, and zileuton. These molecules were selected on the basis of their variety of hepatocellular effects observed such as necrosis, cholestasis, steatosis, and induction of CYP P450 enzymes. We used a low-density DNA microarray containing 59 genes chosen as relevant toxic and metabolic markers. The in vitro gene expression data generated in this study were generally in good agreement with the literature, which mainly concerns in vivo data. Furthermore, gene expression profiles observed in this study have been confirmed for several genes by real-time PCR assays. All the tested drugs generated a specific gene expression profile. Our results show that even with a relatively limited gene set, gene expression profiling allows a certain degree of classification of compounds with similar hepatocellular toxicities such as cholestasis, necrosis. The clustering analysis revealed that the compounds known to cause steatosis were linked, suggesting that they functionally regulate similar genes and possibly act through the same mechanisms of action. On the other hand, the drugs inducing necrosis and cholestasis were pooled in the same cluster. The drugs arbitrarily classified as the CYP450 inducers formed individual clusters. In conclusion, this study suggests that low-density microarrays could be useful in toxicological studies.

Evaluation of drug transporters' significance for multidrug resistance in head and neck squamous cell carcinoma.

BACKGROUND: Multidrug resistance (MDR) hampers chemotherapy in head and neck squamous cell carcinoma (HNSCC). There is little information about MDR mediating drug transporters in HNSCC. METHODS: Nine HNSCC cell lines were characterized concerning drug transporter expression and susceptibility to cisplatin, paclitaxel, and 5-fluorouracil (5-FU) using a DNA microarray and proliferation assays. Three cells lines were precisely investigated for transporter expression using reverse transcriptase-polymerase chain reaction (RT-PCR) and drug resistance before and after simulated chemotherapies. RESULTS: The cell lines differed in drug resistance and transporter expression. Cisplatin and paclitaxel resistances were inversely associated. Drugs rendered HNSCC cells resistant, cross-resistant, or cross-sensitive. Sensitivity changes were accompanied by transporter mRNA expression changes. CONCLUSION: Paclitaxel was identified as a potent inducer of numerous drug transporters and phenotypic MDR in HNSCC. Thus, cytostatic susceptibility of HNSCC cannot exclusively be deduced from the expression of single transporters but more likely of the entire transportome and non-transporter modulators.

Microarray Expression Profiling of ABC Transporters in Human Breast Cancer.

The ATP-binding cassette (ABC) transporters are highly conserved genes involved in the translocation of molecules through biological membranes. Several of them are involved in tumor drug resistance, and it is thought that many others may contribute to the development of the tumor phenotype in a still unknown way. A low-density DNA microarray was recently developed for the analysis of 38 ABC-transporter genes and 3 other transporters. In the present pilot study, clinical samples from 16 breast cancer patients were tested. Of the 41 transporters analyzed, 10 were not or very seldom expressed, while 23 were found to be expressed, sometimes at very high levels, in the majority of the tumors. Comparison of the treated and untreated tumors showed an unexpected similarity of results. The signal obtained on the capture probes for ABCC6/8/9 was, however, found to be higher in the treated samples. The microarray data were validated on 15 ABC-transporter genes by real-time PCR. The present results showed that the expression of the majority of the ABC transporters was a clear feature of breast tumors, whether treated or not.

CREB activation induced by mitochondrial dysfunction triggers triglyceride accumulation in 3T3-L1 preadipocytes.

Several mitochondrial pathologies are characterized by lipid redistribution and microvesicular cell phenotypes resulting from triglyceride accumulation in lipid-metabolizing tissues. However, the molecular mechanisms underlying abnormal fat distribution induced by mitochondrial dysfunction remain poorly understood. In this study, we show that inhibition of respiratory complex III by antimycin A as well as inhibition of mitochondrial protein synthesis trigger the accumulation of triglyceride vesicles in 3T3-L1 fibroblasts. We also show that treatment with antimycin A triggers CREB activation in these cells. To better delineate how mitochondrial dysfunction induces triglyceride accumulation in preadipocytes, we developed a low-density DNA microarray containing 89 probes, which allows gene expression analysis for major effectors and/or markers of adipogenesis. We thus determined gene expression profiles in 3T3-L1 cells incubated with antimycin A and compared the patterns obtained with differentially expressed genes during the course of in vitro adipogenesis induced by a standard pro-adipogenic cocktail. After an 8-day treatment, a set of 39 genes was found to be differentially expressed in cells treated with antimycin A, among them CCAAT/enhancer-binding protein alpha (C/EBPalpha), C/EBP homologous protein-10 (CHOP-10), mitochondrial glycerol-3-phosphate dehydrogenase (GPDmit), and stearoyl-CoA desaturase 1 (SCD1). We also demonstrate that overexpression of two dominant negative mutants of the cAMP-response element-binding protein CREB (K-CREB and M1-CREB) and siRNA transfection, which disrupt the factor activity and expression, respectively, inhibit antimycin-A-induced triglyceride accumulation. Furthermore, CREB knockdown with siRNA also downregulates the expression of several genes that contain cAMP-response element (CRE) sites in their promoter, among them one that is potentially involved in synthesis of triglycerides such as SCD1. These results highlight a new role for CREB in the control of triglyceride metabolism during the adaptative response of preadipocytes to mitochondrial dysfunction.

An evaluation of a low-density DNA microarray using cytochrome P450 inducers.

The aim of this study was to validate a low-density DNA microarray "Rat HepatoChip", which contains 59 genes from a range of potential toxic markers and drug metabolism-related genes. Liver mRNA was isolated from rats dosed with six different chemicals, dexamethasone, troleandomycin, miconazole, clotrimazole, and methylclofanapate, which are all known to induce different cytochrome P450 genes, and isoniazid, which does not cause histopathological changes. Replicate microarrays were used to measure the variability in the chips and in the process. The average variability in signal between different chips observed in triplicate experiments was 33% ranging from 21 to 39% depending on genes. We also demonstrated a strong correlation between the liver histopathology and the gene expression profiles indicating that the gene expression profile reflects histopathological changes. These results suggest that the Rat HepatoChip microarray may provide a fast and effective tool for assessing the toxicity profile of developmental drug candidates during the drug discovery process.