Analysis of dose-response effects on gene expression data with comparison of two microarray platforms

Zafirlukast Is a Promising Scaffold for Selectively Inhibiting TNFR1 Signaling

Tumor necrosis factor (TNF) plays an important role in the pathogenesis of inflammatory and autoimmune diseases such as rheumatoid arthritis and Crohn's disease. The biological effects of TNF are mediated by binding to TNF receptors, TNF receptor 1 (TNFR1), or TNF receptor 2 (TNFR2), and this coupling makes TNFR1-specific inhibition by small-molecule therapies essential to avoid deleterious side effects. Recently, we engineered a time-resolved fluorescence resonance energy transfer biosensor for high-throughput screening of small molecules that modulate TNFR1 conformational states and identified zafirlukast as a compound that inhibits receptor activation, albeit at low potency. Here, we synthesized 16 analogues of zafirlukast and tested their potency and specificity for TNFR1 signaling. Using cell-based functional assays, we identified three analogues with significantly improved efficacy and potency, each of which induces a conformational change in the receptor (as measured by fluorescence resonance energy transfer (FRET) in cells). The best analogue decreased NF-κB activation by 2.2-fold, IκBα efficiency by 3.3-fold, and relative potency by two orders of magnitude. Importantly, we showed that the analogues do not block TNF binding to TNFR1 and that binding to the receptor's extracellular domain is strongly cooperative. Despite these improvements, the best candidate's maximum inhibition of NF-κB is only 63%, leaving room for further improvements to the zafirlukast scaffold to achieve full inhibition and prove its potential as a therapeutic lead. Interestingly, while we find that the analogues also bind to TNFR2 in vitro, they do not inhibit TNFR2 function in cells or cause any conformational changes upon binding. Thus, these lead compounds should also be used as reagents to study conformational-dependent activation of TNF receptors.

BMDx: a graphical Shiny application to perform Benchmark Dose analysis for transcriptomics data

MOTIVATION

The analysis of dose-dependent effects on the gene expression is gaining attention in the field of toxicogenomics. Currently available computational methods are usually limited to specific omics platforms or biological annotations and are able to analyse only one experiment at a time.

RESULTS

We developed the software BMDx with a graphical user interface for the Benchmark Dose (BMD) analysis of transcriptomics data. We implemented an approach based on the fitting of multiple models and the selection of the optimal model based on the Akaike Information Criterion. The BMDx tool takes as an input a gene expression matrix and a phenotype table, computes the BMD, its related values, and IC50/EC50 estimations. It reports interactive tables and plots that the user can investigate for further details of the fitting, dose effects and functional enrichment. BMDx allows a fast and convenient comparison of the BMD values of a transcriptomics experiment at different time points and an effortless way to interpret the results. Furthermore, BMDx allows to analyse and to compare multiple experiments at once.

AVAILABILITY AND IMPLEMENTATION

BMDx is implemented as an R/Shiny software and is available at https://github.com/Greco-Lab/BMDx/.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Gene-signature-derived IC50s/EC50s reflect the potency of causative upstream targets and downstream phenotypes

Multiplexed gene-signature-based phenotypic assays are increasingly used for the identification and profiling of small molecule-tool compounds and drugs. Here we introduce a method (provided as R-package) for the quantification of the dose-response potency of a gene-signature as EC₅₀ and IC₅₀ values. Two signaling pathways were used as models to validate our methods: beta-adrenergic agonistic activity on cAMP generation (dedicated dataset generated for this study) and EGFR inhibitory effect on cancer cell viability. In both cases, potencies derived from multi-gene expression data were highly correlated with orthogonal potencies derived from cAMP and cell growth readouts, and superior to potencies derived from single individual genes. Based on our results we propose gene-signature potencies as a novel valid alternative for the quantitative prioritization, optimization and development of novel drugs.

Transcriptomics in Toxicogenomics, Part III: Data Modelling for Risk Assessment

Transcriptomics data are relevant to address a number of challenges in Toxicogenomics (TGx). After careful planning of exposure conditions and data preprocessing, the TGx data can be used in predictive toxicology, where more advanced modelling techniques are applied. The large volume of molecular profiles produced by omics-based technologies allows the development and application of artificial intelligence (AI) methods in TGx. Indeed, the publicly available omics datasets are constantly increasing together with a plethora of different methods that are made available to facilitate their analysis, interpretation and the generation of accurate and stable predictive models. In this review, we present the state-of-the-art of data modelling applied to transcriptomics data in TGx. We show how the benchmark dose (BMD) analysis can be applied to TGx data. We review read across and adverse outcome pathways (AOP) modelling methodologies. We discuss how network-based approaches can be successfully employed to clarify the mechanism of action (MOA) or specific biomarkers of exposure. We also describe the main AI methodologies applied to TGx data to create predictive classification and regression models and we address current challenges. Finally, we present a short description of deep learning (DL) and data integration methodologies applied in these contexts. Modelling of TGx data represents a valuable tool for more accurate chemical safety assessment. This review is the third part of a three-article series on Transcriptomics in Toxicogenomics.

Distinct gene expression changes in human peripheral blood mononuclear cells treated with different tobacco product preparations

Cigarette smoking exerts diverse physiological effects including immune suppression. To better characterize the biological effects of different categories of tobacco products, a genome-wide gene expression study was performed. Transcriptomic profiling was performed in PBMCs treated with different equi-nicotine units of aqueous extracts of cigarette smoke (termed Whole Smoke-Conditioned Medium, or WS-CM), or a single dose smokeless tobacco extract (STE) prepared from reference tobacco products. WS-CM induced dose-dependent changes in the expression of several genes. No significant expression differences between low WS-CM and media control were detected. However, transcripts were significantly affected by medium WS-CM (479), high WS-CM (2, 703), and STE (2, 156). The overlap between medium WS-CM and STE, and high WS-CM and STE, was minimal (34 and 65 transcripts, respectively). Hierarchical clustering revealed that gene expression profiles for STE and medium WS-CM co-clustered, while those affected by the high dose of WS-CM clustered distinctly. Functional analysis revealed that WS-CM, but not STE, uniquely affected genes involved in immune cell development and inflammatory response. Cascades of upstream regulators (e.g., TNF, IL1β, NFƙB) were identified for the observed gene expression changes and generally suppressed by WS-CM, but not by STE. Collectively, these findings demonstrate that combustible and non-combustible tobacco products elicit distinct biological effects, which could explain the observed chronic immune suppression in smokers.

Multiple confidence intervals for selected parameters adjusted for the false coverage rate in monotone dose-response microarray experiments

Benjamini and Yekutieli () introduced the concept of the false coverage-statement rate (FCR) to account for selection when the confidence intervals (CIs) are constructed only for the selected parameters. Dose-response analysis in dose-response microarray experiments is conducted only for genes having monotone dose-response relationship, which is a selection problem. In this paper, we consider multiple CIs for the mean gene expression difference between the highest dose and control in monotone dose-response microarray experiments for selected parameters adjusted for the FCR. A simulation study is conducted to study the performance of the method proposed. The method is applied to a real dose-response microarray experiment with 16, 998 genes for illustration.

Testing differentially expressed genes in dose-response studies and with ordinal phenotypes

When testing for differentially expressed genes between more than two groups, the groups are often defined by dose levels in dose-response experiments or ordinal phenotypes, such as disease stages. We discuss the potential of a new approach that uses the levels’ ordering without making any structural assumptions, such as monotonicity, by testing for zero variance components in a mixed models framework. Since the mixed effects model approach borrows strength across doses/levels, the test proposed can also be applied when the number of dose levels/phenotypes is large and/or the number of subjects per group is small. We illustrate the new test in simulation studies and on several publicly available datasets and compare it to alternative testing procedures. All tests considered are implemented in R and are publicly available. The new approach offers a very fast and powerful way to test for differentially expressed genes between ordered groups without making restrictive assumptions with respect to the true relationship between factor levels and response.

Isotonic Regression Based-Method in Quantitative High-Throughput Screenings for Genotoxicity

Quantitative high-throughput screenings (qHTSs) for genotoxicity are conducted as part of comprehensive toxicology screening projects. The most widely used method is to compare the dose-response data of a wild-type and DNA repair gene knockout mutants, using model-fitting to the Hill equation (HE). However, this method performs poorly when the observed viability does not fit the equation well, as frequently happens in qHTS. More capable methods must be developed for qHTS where large data variations are unavoidable. In this study, we applied an isotonic regression (IR) method and compared its performance with HE under multiple data conditions. When dose-response data were suitable to draw HE curves with upper and lower asymptotes and experimental random errors were small, HE was better than IR, but when random errors were big, there was no difference between HE and IR. However, when the drawn curves did not have two asymptotes, IR showed better performance (p < 0.05, exact paired Wilcoxon test) with higher specificity (65% in HE vs. 96% in IR). In summary, IR performed similarly to HE when dose-response data were optimal, whereas IR clearly performed better in suboptimal conditions. These findings indicate that IR would be useful in qHTS for comparing dose-response data.

Cross-platform comparison of microarray data using order restricted inference

MOTIVATION

Titration experiments measuring the gene expression from two different tissues, along with total RNA mixtures of the pure samples, are frequently used for quality evaluation of microarray technologies. Such a design implies that the true mRNA expression of each gene, is either constant or follows a monotonic trend between the mixtures, applying itself to the use of order restricted inference procedures. Exploiting only the postulated monotonicity of titration designs, we propose three statistical analysis methods for the validation of high-throughput genetic data and corresponding preprocessing techniques.

RESULTS

Our methods allow for inference of accuracy, repeatability and cross-platform agreement, with minimal required assumptions regarding the underlying data generating process. Therefore, they are readily applicable to all sorts of genetic high-throughput data independent of the degree of preprocessing. An application to the EMERALD dataset was used to demonstrate how our methods provide a rich spectrum of easily interpretable quality metrics and allow the comparison of different microarray technologies and normalization methods. The results are on par with previous work, but provide additional new insights that cast doubt on the utility of popular preprocessing techniques, specifically concerning the EMERALD projects dataset.

AVAILABILITY

All datasets are available on EBI's ArrayExpress web site http://www.ebi.ac.uk/microarray-as/ae/) under accession numbers E-TABM-536, E-TABM-554 and E-TABM-555. Source code implemented in C and R is available at: http://statistics.msi.meduniwien.ac.at/float/cross_platform/. Methods for testing and variance decomposition have been made available in the R-package orQA, which can be downloaded and installed from CRAN http://cran.r-project.org.

Identification of gene signatures and molecular markers for human lung cancer prognosis using an in vitro lung carcinogenesis system

Lung cancer continues to be a major deadly malignancy. The mortality of this disease could be reduced by improving the ability to predict cancer patients' survival. We hypothesized that genes differentially expressed among cells constituting an in vitro human lung carcinogenesis model consisting of normal, immortalized, transformed, and tumorigenic bronchial epithelial cells are relevant to the clinical outcome of non-small cell lung cancer (NSCLC). Multidimensional scaling, microarray, and functional pathways analyses of the transcriptomes of the above cells were done and combined with integrative genomics to incorporate the microarray data with published NSCLC data sets. Up-regulated (n = 301) and down-regulated genes (n = 358) displayed expression level variation across the in vitro model with progressive changes in cancer-related molecular functions. A subset of these genes (n = 584) separated lung adenocarcinoma clinical samples (n = 361) into two clusters with significant survival differences. Six genes, UBE2C, TPX2, MCM2, MCM6, FEN1, and SFN, selected by functional array analysis, were also effective in prognosis. The mRNA and protein levels of one these genes-UBE2C-were significantly up-regulated in NSCLC tissue relative to normal lung and increased progressively in lung lesions. Moreover, stage I NSCLC patients with positive UBE2C expression exhibited significantly poorer overall and progression-free survival than patients with negative expression. Our studies with this in vitro model have lead to the identification of a robust six-gene signature, which may be valuable for predicting the survival of lung adenocarcinoma patients. Moreover, one of those genes, UBE2C, seems to be a powerful biomarker for NSCLC survival prediction.

Transcriptional response of rat frontal cortex following acute in vivo exposure to the pyrethroid insecticides permethrin and deltamethrin

Background

Pyrethroids are neurotoxic pesticides that interact with membrane bound ion channels in neurons and disrupt nerve function. The purpose of this study was to characterize and explore changes in gene expression that occur in the rat frontal cortex, an area of CNS affected by pyrethroids, following an acute low-dose exposure.

Results

Rats were acutely exposed to either deltamethrin (0.3 – 3 mg/kg) or permethrin (1 – 100 mg/kg) followed by collection of cortical tissue at 6 hours. The doses used range from those that cause minimal signs of intoxication at the behavioral level to doses well below apparent no effect levels in the whole animal. A statistical framework based on parallel linear (SAM) and isotonic regression (PIR) methods identified 95 and 53 probe sets as dose-responsive. The PIR analysis was most sensitive for detecting transcripts with changes in expression at the NOAEL dose. A sub-set of genes (Camk1g, Ddc, Gpd3, c-fos and Egr1) was then confirmed by qRT-PCR and examined in a time course study. Changes in mRNA levels were typically less than 3-fold in magnitude across all components of the study. The responses observed are consistent with pyrethroids producing increased neuronal excitation in the cortex following a low-dose in vivo exposure. In addition, Significance Analysis of Function and Expression (SAFE) identified significantly enriched gene categories common for both pyrethroids, including some relating to branching morphogenesis. Exposure of primary cortical cell cultures to both compounds resulted in an increase (~25%) in the number of neurite branch points, supporting the results of the SAFE analysis.

Conclusion

In the present study, pyrethroids induced changes in gene expression in the frontal cortex near the threshold for decreases in ambulatory motor activity in vivo. The penalized regression methods performed similarly in detecting dose-dependent changes in gene transcription. Finally, SAFE analysis of gene expression data identified branching morphogenesis as a biological process sensitive to pyrethroids and subsequent in vitro experiments confirmed this predicted effect. The novel findings regarding pyrethroid effects on branching morphogenesis indicate these compounds may act as developmental neurotoxicants that affect normal neuronal morphology.

Assessing the information content of short time series expression data

Due to experimental constraints, the sampling of biological system with microarray data is severely constrained. In similar fashion to sampling theory of signals, the under-sampling of a system oftentimes leads to sub-optimal results from which it is difficult to draw proper conclusions. In our work we create a mathematical framework which will show that the sampling methodology for short time series microarray data may lead to data whose ability to distinguish non-random behavior within the biological system is severely constrained.

Toxicogenomic analysis incorporating operon-transcriptional coupling and toxicant concentration-expression response: analysis of MX-treated Salmonella

Background

Deficiencies in microarray technology cause unwanted variation in the hybridization signal, obscuring the true measurements of intracellular transcript levels. Here we describe a general method that can improve microarray analysis of toxicant-exposed cells that uses the intrinsic power of transcriptional coupling and toxicant concentration-expression response data. To illustrate this approach, we characterized changes in global gene expression induced in Salmonella typhimurium TA100 by 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone (MX), the primary mutagen in chlorinated drinking water. We used the co-expression of genes within an operon and the monotonic increases or decreases in gene expression relative to increasing toxicant concentration to augment our identification of differentially expressed genes beyond Bayesian-t analysis.

Results

Operon analysis increased the number of altered genes by 95% from the list identified by a Bayesian t-test of control to the highest concentration of MX. Monotonic analysis added 46% more genes. A functional analysis of the resulting 448 differentially expressed genes yielded functional changes beyond what would be expected from only the mutagenic properties of MX. In addition to gene-expression changes in DNA-damage response, MX induced changes in expression of genes involved in membrane transport and porphyrin metabolism, among other biological processes. The disruption of porphyrin metabolism might be attributable to the structural similarity of MX, which is a chlorinated furanone, to ligands indigenous to the porphyrin metabolism pathway. Interestingly, our results indicate that the lexA regulon in Salmonella, which partially mediates the response to DNA damage, may contain only 60% of the genes present in this regulon in E. coli. In addition, nanH was found to be highly induced by MX and contains a putative lexA regulatory motif in its regulatory region, suggesting that it may be regulated by lexA.

Conclusion

Operon and monotonic analyses improved the determination of differentially expressed genes beyond that of Bayesian-t analysis, showing that MX alters cellular metabolism involving pathways other than DNA damage. Because co-expression of similarly functioning genes also occurs in eukaryotes, this method has general applicability for improving analysis of toxicogenomic data.

Novel Microarray-Based Method for Estimating Exposure to Ionizing Radiation

Accurate estimation of the dose of ionizing radiation to which individuals have been exposed is critical for therapeutic treatment. We investigated whether gene expression profiles could be used to evaluate the dose received, thereby serving as a biological dosimeter. We used cDNA microarrays to monitor changes in gene expression profiles induced by ionizing radiation in mouse total blood. The subsets of genes best characterizing each dose were identified by resampling the original data set and calculating the intersection of the dose signatures. This analytical strategy minimizes the impact of potential genetic/epigenetic variation between mice and overcomes the bias in gene selection inherent to microarray technology. The significance of the identified signatures was evaluated by monitoring the type I error rate by in silico negative control simulation. Based on the distribution of the mean ratios of the selected probes, we were able to identify transcription profiles giving 83% to 100% correct estimation of the dose received by test mice, demonstrating that the selected probes could be used to determine the dose of radiation to which the animals had been exposed. This method could potentially be generalized to determine the level of exposure to other toxins and could be used to develop new related clinical applications.

An open-access long oligonucleotide microarray resource for analysis of the human and mouse transcriptomes

Two collections of oligonucleotides have been designed for preparing pangenomic human and mouse microarrays. A total of 148 993 and 121 703 oligonucleotides were designed against human and mouse transcripts. Quality scores were created in order to select 25 342 human and 24 109 mouse oligonucleotides. They correspond to: (i) a BLAST-specificity score; (ii) the number of expressed sequence tags matching each probe; (iii) the distance to the 3′ end of the target mRNA. Scores were also used to compare in silico the two microarrays with commercial microarrays. The sets described here, called RNG/MRC collections, appear at least as specific and sensitive as those from the commercial platforms. The RNG/MRC collections have now been used by an Anglo-French consortium to distribute more than 3500 microarrays to the academic community. Ad hoc identification of tissue-specific transcripts and a ∼80% correlation with hybridizations performed on Affymetrix GeneChip™ suggest that the RNG/MRC microarrays perform well. This work provides a comprehensive open resource for investigators working on human and mouse transcriptomes, as well as a generic method to generate new microarray collections in other organisms. All information related to these probes, as well as additional information about commercial microarrays have been stored in a freely-accessible database called MEDIANTE.