On Methods for Gene Function Scoring as a Means of Facilitating the Interpretation of Microarray Results

TMEA: A Thermodynamically Motivated Framework for Functional Characterization of Biological Responses to System Acclimation

The objective of gene set enrichment analysis (GSEA) in modern biological studies is to identify functional profiles in huge sets of biomolecules generated by high-throughput measurements of genes, transcripts, metabolites, and proteins. GSEA is based on a two-stage process using classical statistical analysis to score the input data and subsequent testing for overrepresentation of the enrichment score within a given functional coherent set. However, enrichment scores computed by different methods are merely statistically motivated and often elusive to direct biological interpretation. Here, we propose a novel approach, called Thermodynamically Motivated Enrichment Analysis (TMEA), to account for the energy investment in biological relevant processes. Therefore, TMEA is based on surprisal analysis, which offers a thermodynamic-free energy-based representation of the biological steady state and of the biological change. The contribution of each biomolecule underlying the changes in free energy is used in a Monte Carlo resampling procedure resulting in a functional characterization directly coupled to the thermodynamic characterization of biological responses to system perturbations. To illustrate the utility of our method on real experimental data, we benchmark our approach on plant acclimation to high light and compare the performance of TMEA with the most frequently used method for GSEA.

Predictive modelling using pathway scores: robustness and significance of pathway collections

BACKGROUND

Transcriptomic data is often used to build statistical models which are predictive of a given phenotype, such as disease status. Genes work together in pathways and it is widely thought that pathway representations will be more robust to noise in the gene expression levels. We aimed to test this hypothesis by constructing models based on either genes alone, or based on sample specific scores for each pathway, thus transforming the data to a 'pathway space'. We progressively degraded the raw data by addition of noise and examined the ability of the models to maintain predictivity.

RESULTS

Models in the pathway space indeed had higher predictive robustness than models in the gene space. This result was independent of the workflow, parameters, classifier and data set used. Surprisingly, randomised pathway mappings produced models of similar accuracy and robustness to true mappings, suggesting that the success of pathway space models is not conferred by the specific definitions of the pathway. Instead, predictive models built on the true pathway mappings led to prediction rules with fewer influential pathways than those built on randomised pathways. The extent of this effect was used to differentiate pathway collections coming from a variety of widely used pathway databases.

CONCLUSIONS

Prediction models based on pathway scores are more robust to degradation of gene expression information than the equivalent models based on ungrouped genes. While models based on true pathway scores are not more robust or accurate than those based on randomised pathways, true pathways produced simpler prediction rules, emphasizing a smaller number of pathways.

Avoiding the pitfalls of gene set enrichment analysis with SetRank

Background

The purpose of gene set enrichment analysis (GSEA) is to find general trends in the huge lists of genes or proteins generated by many functional genomics techniques and bioinformatics analyses.

Results

Here we present SetRank, an advanced GSEA algorithm which is able to eliminate many false positive hits. The key principle of the algorithm is that it discards gene sets that have initially been flagged as significant, if their significance is only due to the overlap with another gene set. The algorithm is explained in detail and its performance is compared to that of other methods using objective benchmarking criteria. Furthermore, we explore how sample source bias can affect the results of a GSEA analysis.

Conclusions

The benchmarking results show that SetRank is a highly specific tool for GSEA. Furthermore, we show that the reliability of results can be improved by taking sample source bias into account. SetRank is available as an R package and through an online web interface.

Chronic drug-induced effects on contractile motion properties and cardiac biomarkers in human induced pluripotent stem cell-derived cardiomyocytes

BACKGROUND AND PURPOSE

In the pharmaceutical industry risk assessments of chronic cardiac safety liabilities are mostly performed during late stages of preclinical drug development using in vivo animal models. Here, we explored the potential of human induced pluripotent stem cell-derived cardiomyocytes (hiPS-CMs) to detect chronic cardiac risks such as drug-induced cardiomyocyte toxicity.

EXPERIMENTAL APPROACH

Video microscopy-based motion field imaging was applied to evaluate the chronic effect (over 72 h) of cardiotoxic drugs on the contractile motion of hiPS-CMs. In parallel, the release of cardiac troponin I (cTnI), heart fatty acid binding protein (FABP3) and N-terminal pro-brain natriuretic peptide (NT-proBNP) was analysed from cell medium, and transcriptional profiling of hiPS-CMs was done at the end of the experiment.

KEY RESULTS

Different cardiotoxic drugs altered the contractile motion properties of hiPS-CMs together with increasing the release of cardiac biomarkers. FABP3 and cTnI were shown to be potential surrogates to predict cardiotoxicity in hiPS-CMs, whereas NT-proBNP seemed to be a less valuable biomarker. Furthermore, drug-induced cardiotoxicity produced by chronic exposure of hiPS-CMs to arsenic trioxide, doxorubicin or panobinostat was associated with different profiles of changes in contractile parameters, biomarker release and transcriptional expression.

CONCLUSION AND IMPLICATIONS

We have shown that a parallel assessment of motion field imaging-derived contractile properties, release of biomarkers and transcriptional changes can detect diverse mechanisms of chronic drug-induced cardiac liabilities in hiPS-CMs. Hence, hiPS-CMs could potentially improve and accelerate cardiovascular de-risking of compounds at earlier stages of drug discovery.

LINKED ARTICLES

This article is part of a themed section on New Insights into Cardiotoxicity Caused by Chemotherapeutic Agents. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v174.21/issuetoc.

HCMV Displays a Unique Transcriptome of Immunomodulatory Genes in Primary Monocyte-Derived Cell Types

Human cytomegalovirus (HCMV) is a betaherpesvirus which rarely presents problems in healthy individuals, yet may result in severe morbidity in immunocompromised patients and in immune-naïve neonates. HCMV has a large 235 kb genome with a coding capacity of at least 165 open reading frames (ORFs). This large genome allows complex gene regulation resulting in different sets of transcripts during lytic and latent infection. While latent virus mainly resides within monocytes and CD34+ progenitor cells, reactivation to lytic infection is driven by differentiation towards terminally differentiated myeloid dendritic cells and macrophages. Consequently, it has been suggested that macrophages and dendritic cells contribute to viral spread in vivo. Thus far only limited knowledge is available on the expression of HCMV genes in terminally differentiated myeloid primary cells and whether or not the virus exhibits a different set of lytic genes in primary cells compared with lytic infection in NHDF fibroblasts. To address these questions, we used Illumina next generation sequencing to determine the HCMV transcriptome in macrophages and dendritic cells during lytic infection and compared it to the transcriptome in NHDF fibroblasts. Here, we demonstrate unique expression profiles in macrophages and dendritic cells which significantly differ from the transcriptome in fibroblasts mainly by modulating the expression of viral transcripts involved in immune modulation, cell tropism and viral spread. In a head to head comparison between macrophages and dendritic cells, we observed that factors involved in viral spread and virion composition are differentially regulated suggesting that the plasticity of the virion facilitates the infection of surrounding cells. Taken together, this study provides the full transcript expression analysis of lytic HCMV genes in monocyte-derived type 1 and type 2 macrophages as well as in monocyte-derived dendritic cells. Thereby underlining the potential of HCMV to adapt to or influence different cellular environments to promote its own survival.

Fungus-Derived Neoechinulin B as a Novel Antagonist of Liver X Receptor, Identified by Chemical Genetics Using a Hepatitis C Virus Cell Culture System

Cell culture systems reproducing virus replication can serve as unique models for the discovery of novel bioactive molecules. Here, using a hepatitis C virus (HCV) cell culture system, we identified neoechinulin B (NeoB), a fungus-derived compound, as an inhibitor of the liver X receptor (LXR). NeoB was initially identified by chemical screening as a compound that impeded the production of infectious HCV. Genome-wide transcriptome analysis and reporter assays revealed that NeoB specifically inhibits LXR-mediated transcription. NeoB was also shown to interact directly with LXRs. Analysis of structural analogs suggested that the molecular interaction of NeoB with LXR correlated with the capacity to inactivate LXR-mediated transcription and to modulate lipid metabolism in hepatocytes. Our data strongly suggested that NeoB is a novel LXR antagonist. Analysis using NeoB as a bioprobe revealed that LXRs support HCV replication: LXR inactivation resulted in dispersion of double-membrane vesicles, putative viral replication sites. Indeed, cells treated with NeoB showed decreased replicative permissiveness for poliovirus, which also replicates in double-membrane vesicles, but not for dengue virus, which replicates via a distinct membrane compartment. Together, our data suggest that LXR-mediated transcription regulates the formation of virus-associated membrane compartments. Significantly, inhibition of LXRs by NeoB enhanced the activity of all known classes of anti-HCV agents, and NeoB showed especially strong synergy when combined with interferon or an HCV NS5A inhibitor. Thus, our chemical genetics analysis demonstrates the utility of the HCV cell culture system for identifying novel bioactive molecules and characterizing the virus-host interaction machinery.

IMPORTANCE

Hepatitis C virus (HCV) is highly dependent on host factors for efficient replication. In the present study, we used an HCV cell culture system to screen an uncharacterized chemical library. Our results identified neoechinulin B (NeoB) as a novel inhibitor of the liver X receptor (LXR). NeoB inhibited the induction of LXR-regulated genes and altered lipid metabolism. Intriguingly, our results indicated that LXRs are critical to the process of HCV replication: LXR inactivation by NeoB disrupted double-membrane vesicles, putative sites of viral replication. Moreover, NeoB augmented the antiviral activity of all known classes of currently approved anti-HCV agents without increasing cytotoxicity. Thus, our strategy directly links the identification of novel bioactive compounds to basic virology and the development of new antiviral agents.

MELK-T1, a small-molecule inhibitor of protein kinase MELK, decreases DNA-damage tolerance in proliferating cancer cells

Protein kinase MELK has oncogenic properties and is highly overexpressed in some tumors. In the present study, we show that a novel MELK inhibitor causes both the inhibition and degradation of MELK, culminating in replication stress and a senescence phenotype.

Maternal embryonic leucine zipper kinase (MELK), a serine/threonine protein kinase, has oncogenic properties and is overexpressed in many cancer cells. The oncogenic function of MELK is attributed to its capacity to disable critical cell-cycle checkpoints and reduce replication stress. Most functional studies have relied on the use of siRNA/shRNA-mediated gene silencing. In the present study, we have explored the biological function of MELK using MELK-T1, a novel and selective small-molecule inhibitor. Strikingly, MELK-T1 triggered a rapid and proteasome-dependent degradation of the MELK protein. Treatment of MCF-7 (Michigan Cancer Foundation-7) breast adenocarcinoma cells with MELK-T1 induced the accumulation of stalled replication forks and double-strand breaks that culminated in a replicative senescence phenotype. This phenotype correlated with a rapid and long-lasting ataxia telangiectasia-mutated (ATM) activation and phosphorylation of checkpoint kinase 2 (CHK2). Furthermore, MELK-T1 induced a strong phosphorylation of p53 (cellular tumour antigen p53), a prolonged up-regulation of p21 (cyclin-dependent kinase inhibitor 1) and a down-regulation of FOXM1 (Forkhead Box M1) target genes. Our data indicate that MELK is a key stimulator of proliferation by its ability to increase the threshold for DNA-damage tolerance (DDT). Thus, targeting MELK by the inhibition of both its catalytic activity and its protein stability might sensitize tumours to DNA-damaging agents or radiation therapy by lowering the DNA-damage threshold.

Connecting gene expression data from connectivity map and in silico target predictions for small molecule mechanism-of-action analysis

Integrating gene expression profiles with certain proteins can improve our understanding of the fundamental mechanisms in protein-ligand binding. This paper spotlights the integration of gene expression data and target prediction scores, providing insight into mechanism of action (MoA). Compounds are clustered based upon the similarity of their predicted protein targets and each cluster is linked to gene sets using Linear Models for Microarray Data. MLP analysis is used to generate gene sets based upon their biological processes and a qualitative search is performed on the homogeneous target-based compound clusters to identify pathways. Genes and proteins were linked through pathways for 6 of the 8 MCF7 and 6 of the 11 PC3 clusters. Three compound clusters are studied; (i) the target-driven cluster involving HSP90 inhibitors, geldanamycin and tanespimycin induces differential expression for HSP90-related genes and overlap with pathway response to unfolded protein. Gene expression results are in agreement with target prediction and pathway annotations add information to enable understanding of MoA. (ii) The antipsychotic cluster shows differential expression for genes LDLR and INSIG-1 and is predicted to target CYP2D6. Pathway steroid metabolic process links the protein and respective genes, hypothesizing the MoA for antipsychotics. A sub-cluster (verepamil and dexverepamil), although sharing similar protein targets with the antipsychotic drug cluster, has a lower intensity of expression profile on related genes, indicating that this method distinguishes close sub-clusters and suggests differences in their MoA. Lastly, (iii) the thiazolidinediones drug cluster predicted peroxisome proliferator activated receptor (PPAR) PPAR-alpha, PPAR-gamma, acyl CoA desaturase and significant differential expression of genes ANGPTL4, FABP4 and PRKCD. The targets and genes are linked via PPAR signalling pathway and induction of apoptosis, generating a hypothesis for the MoA of thiazolidinediones. Our analysis show one or more underlying MoA for compounds and were well-substantiated with literature.

How advancement in biological network analysis methods empowers proteomics

Proteomics provides important information--that may not be inferable from indirect sources such as RNA or DNA--on key players in biological systems or disease states. However, it suffers from coverage and consistency problems. The advent of network-based analysis methods can help in overcoming these problems but requires careful application and interpretation. This review considers briefly current trends in proteomics technologies and understanding the causes of critical issues that need to be addressed--i.e., incomplete data coverage and inter-sample inconsistency. On the coverage issue, we argue that holistic analysis based on biological networks provides a suitable background on which more robust models and interpretations can be built upon; and we introduce some recently developed approaches. On consistency, group-based approaches based on identified clusters, as well as on properly integrated pathway databases, are particularly useful. Despite that protein interactions and pathway networks are still largely incomplete, given proper quality checks, applications and reasonably sized data sets, they yield valuable insights that greatly complement data generated from quantitative proteomics.

Rare variant collapsing in conjunction with mean log p-value and gradient boosting approaches applied to Genetic Analysis Workshop 17 data

In addition to methods that can identify common variants associated with susceptibility to common diseases, there has been increasing interest in approaches that can identify rare genetic variants. We use the simulated data provided to the participants of Genetic Analysis Workshop 17 (GAW17) to identify both rare and common single-nucleotide polymorphisms and pathways associated with disease status. We apply a rare variant collapsing approach and the usual association tests for common variants to identify candidates for further analysis using pathway-based and tree-based ensemble approaches. We use the mean log p-value approach to identify a top set of pathways and compare it to those used in simulation of GAW17 dataset. We conclude that the mean log p-value approach is able to identify those pathways in the top list and also related pathways. We also use the stochastic gradient boosting approach for the selected subset of single-nucleotide polymorphisms. When compared the result of this tree-based method with the list of single-nucleotide polymorphisms used in dataset simulation, in addition to correct SNPs we observe number of false positives.

The high-level similarity of some disparate gene expression measures

Probe-level data from Affymetrix GeneChips can be summarized in many ways to produce probe-set level gene expression measures (GEMs). Disturbingly, the different approaches not only generate quite different measures but they could also yield very different analysis results. Here, we explore the question of how much the analysis results really do differ, first at the gene level, then at the biological process level. We demonstrate that, even though the gene level results may not necessarily match each other particularly well, as long as there is reasonably strong differentiation between the groups in the data, the various GEMs do in fact produce results that are similar to one another at the biological process level. Not only that the results are biologically relevant. As the extent of differentiation drops, the degree of concurrence weakens, although the biological relevance of findings at the biological process level may yet remain.

The metabolic response of heterotrophic Arabidopsis cells to oxidative stress

To cope with oxidative stress, the metabolic network of plant cells must be reconfigured either to bypass damaged enzymes or to support adaptive responses. To characterize the dynamics of metabolic change during oxidative stress, heterotrophic Arabidopsis (Arabidopsis thaliana) cells were treated with menadione and changes in metabolite abundance and (13)C-labeling kinetics were quantified in a time series of samples taken over a 6 h period. Oxidative stress had a profound effect on the central metabolic pathways with extensive metabolic inhibition radiating from the tricarboxylic acid cycle and including large sectors of amino acid metabolism. Sequential accumulation of metabolites in specific pathways indicated a subsequent backing up of glycolysis and a diversion of carbon into the oxidative pentose phosphate pathway. Microarray analysis revealed a coordinated transcriptomic response that represents an emergency coping strategy allowing the cell to survive the metabolic hiatus. Rather than attempt to replace inhibited enzymes, transcripts encoding these enzymes are in fact down-regulated while an antioxidant defense response is mounted. In addition, a major switch from anabolic to catabolic metabolism is signaled. Metabolism is also reconfigured to bypass damaged steps (e.g. induction of an external NADH dehydrogenase of the mitochondrial respiratory chain). The overall metabolic response of Arabidopsis cells to oxidative stress is remarkably similar to the superoxide and hydrogen peroxide stimulons of bacteria and yeast (Saccharomyces cerevisiae), suggesting that the stress regulatory and signaling pathways of plants and microbes may share common elements.