In recent years, the pharmaceutical industry has placed increasing emphasis on mechanistic characterization of target-related and off target toxicity early in discovery programs. Due to the high attrition rate of drugs related to safety issues in non-clinical and clinical studies, discovery toxicology focus has expanded from beyond the traditional .gold standard. genotoxicity and in vivo safety studies to include early screening assays and alternative models. In addition, more diverse methods have been employed, including molecular, cellular, and computational approaches to better identify, characterize, and predict translational outcomes. In order to identify in vitro endpoints and models with high in vivo predictive value, we have employed a toxicogenomic approach combined with a systems biology, pathway-focused analytical methodology to assist in identification of in vivo-in vitro conserved mechanisms. In one example, we compared gene expression patterns between rat liver (in vivo) and rat primary hepatocytes (in vitro) following treatment with either acetaminophen or naphthylisothiocyanate (ANIT) and characterized toxicity pathways regulated in similar manners in both model systems. Additionally, we compared global expression patterns and pathway perturbations between control rat livers and untreated primary hepatocytes in order to identify significantly perturbed pathways resulting from the process of obtaining and plating primary cells. From this study, we were able to provide mechanistic support for utilization of an in-house rat primary hepatocyte reactive metabolite cytotoxicity assay with in vivo predictive concordance.

Gene expression signatures of toxicity and clinical response benefit both safety assessment and clinical practice. However, difficulties in understanding the association of the signatures to the predicted endpoints have limited their application. Lead by FDA, the Microarray Quality Control (MAQC) II MAQCII project generated 262 signatures for ten clinical and three toxicological endpoints from six gene expression datasets. A comprehensive functional analysis of these signatures and their non-redundant unions was conducted using ontology enrichment, biological network building and interactome connectivity analyses. Different signatures for a given endpoint were more similar at the level of biological entities and transcriptional control than at the gene level. Signatures tended to be enriched in function and pathway in an endpoint and model-specific manner, and showed a topological bias for incoming interactions. Importantly, the level of biological similarity between different signatures correlated positively with the average accuracy of the signature predictions. These findings have implications for the design, analysis, understanding and application of predictive genomics.

The advent of platform -omics technologies has resulted in a drastic increase in biological data covering genetics, RNA/protein expression, metabolomics and protein interactions. It has also driven more open approaches to understanding biological mechanisms, as evidenced by the rise of systems biology. Such open, non-reductionist systems based approaches to scientific discovery require significant experimental input and generate significant amounts of complex data. To run such ambitious experiments on a regular basis is not feasible because of the costs involved in generating the data, analysing the results, constructing hypotheses, and testing these experimentally. In some cases, in silico methods, coupled to extensive data mining, can be used to construct testable hypotheses that emulate the output of systems approaches. Here we will discuss how we have used the GeneGo cystic fibrosis disease pathway maps together with publicly available platform data and literature searching to identify subsets of targets and compounds that would potentially elicit desired cellular responses in a CFTR corrector assay. This approach gave an improved .hit. rate and could potentially identify new indications for existing assets.

Pathway analysis tools assist in the integration and interpretation of the mountain of data generated by omics technologies. Here we present two studies that apply transcriptomics and GeneGo’s MetaCore to assess the toxicological response of the rat kidney. The first study utilized global gene expression profiling to compare male and female Han Wistar, Sprague Dawley, and Fisher rat strains in a model of cisplatin induced renal injury. Pearson Correlation using a list of significantly changed genes from all experimental groups shows clustering influenced in order by gender, control vs. cisplatin treatment, and strain. Gene ontology enrichments for statistically significant changed genes indicate varying effects on processes involved in cell-cycle progression, DNA damage and apoptosis related to the pharmacological effects of cisplatin. Strain and gender transcript differences in response to cisplatin were further characterized by investigating the effects on transporters, xenobiotic metabolism genes and a panel of renal toxicity biomarkers. The second study applies kidney transcriptomic profiling to support a discovery-phase project. Male Han Wistar rats were dosed with 3 different compounds in the same chemical class. With no differentiating pathology, ontology enrichment analysis was able to differentiate the transcript profiles among the 3 compounds. Compounds A and B enrichments show similar ontology distributions implicating genes involved in Phase II xenobiotic metabolism. Compound C enrichments show ontologies related to drug transport and cell-cycle progression. Also, renal toxicity biomarker assays show induction of Kim1 in rats dosed with Compound C. These analyses enabled rank ordering of compounds for renal toxicity liability and supported the selection of Compound A for further development.

With the advent of human geneome sequencing, using sequence information has become another for approach disease solutions. Here we present how to use sequence data, conduct the appropriate statistic and quality assessments and follow through with biological/disease- context solutions. We will specifically use neuronal stem cells at difference states of differentiation to determine specific qualities for each cell type for neuronal therapies.

To become fully immortalised, cells must bypass the permanent cell cycle arrest known as senescence. This is usually achieved through the activation of one of the two known telomere maintenance mechanisms (TMM): telomerase or the alternative lengthening of telomeres (ALT). How the decision between the two mechanisms is made and maintained is currently poorly understood. Gene-expression microarrays of immortalised cell lines and mesenchymal tumour biopsies, previously characterised for telomere maintenance mechanism, were performed. Bioinformatic analysis revealed gene expression patterns suggestive of a potential mesenchymal stem cell origin for ALT. Through the use of MetaCore network building algorithms a complex, multi-layered network of telomerase regulatory mechanisms were revealed. Furthermore, this analysis highlighted a large number of expression differences in c-Myc downstream targets when comparing ALT and telomerase cells. As c-Myc expression itself was not altered the hypothesis that c-Myc activity may be altered in ALT cells at a post-transcriptional level was formulated. After confirming decreased c-Myc activity in ALT cells the study went on to highlight a potential mechanism of c-Myc regulation through TCEAL7 in ALT previously uncharacterized. This study therefore demonstrates the ability of MetaCore to explore complex data sets, drive hypothesis building and direct secondary experimental research.

At the recent GeneGo UGM, I presented some of my thoughts and observations on the presence of a simple oscillatory motif in a number of GeneGo networks. I am currently studying this issue in relation to network interpretation and drug target evaluation. This presentation will expand on those thoughts, especially with respect to coupled oscillators, general physiology, and sleep. I will also attempt to show the close association between my studies of the Feldenkrais Method and applied 'Molecular Physiology.'

A number of sources of bioactivity data have been integrated into one consisted repository. A prime focus was put on consistent annotation of the targets of compounds. The resulting repository was put in a rich biological context through the use of MetaBase. This presentation highlights some of the obstacles encountered in integrating disparate biological data sources, and showcases important applications (such as pathway-based screening sets, pathway enrichment for compounds) as a direct result of an integrated data repository.

Treatment with Sirolimus, an mTOR inhibitor, has been shown to be efficacious in the MRL/lpr and the NZB x NZW F1 mouse models of lupus nephritis, indicating a critical role for mTOR pathway in both models. This type of demonstration of efficacy in animal models is usually a pre-requisite for advancement into clinical development. However, efficacy in an animal model often has not translated to the desired activity in the clinic. Therefore a more profound understanding of the mechanistic similarities and differences between various animal models and human diseases is highly desirable.

Transcriptional profiling was performed in the mouse model and genes associated with lupus nephritis identified. Network analysis revealed that many of nephritis genes identified are known to interact with the mTOR pathway, thus explaining the basis of Sirolimus efficacy in this model. The relevance in human lupus of the pathways identified in the mouse model was explored by constructing the mTOR pathway interactome (consisting of proteins that interact with members of the mTOR pathway) and a strong association of this interactome with genes related to human lupus was noted. Our findings implicate the mTOR pathway as a critical contributor to human lupus. This broad pathway based approach to understanding the similarities in, differences between, animal models and human diseases may have broader utility.

ABSTRACT: One unresolved issue in Cystic Fibrosis research is how defective CFTR trafficking and resulting functional loss is linked to the chronic inflammatory state. Large scale experiments investigating protein or gene expression changes due to altered CFTR (DF508) trafficking have produced long lists of changes with no apparent connection to inflammation. Likewise, experiments documenting the effects of bacteria induced inflammation on bronchial epithelial cell gene expression patterns have yielded no insights into CFTR trafficking. Because the studies were independent, on different platforms with different methodologies and had different objectives, the usual computer based methods of combining and analyzing the data could not be implemented. In an attempt to understand the possible interplay between inflammation and protein trafficking, we combined and analyzed the results of several published studies using MetaMiner (CF), a knowledge base data analysis system created by Cystic Fibrosis Foundation Therapeutics and GeneGo, Inc. Numerous connections were established between genes documented to correct DF508 trafficking when expressed in cell lines (Trzcinska-Danelut, et. al. 2009) and a list of genes differentially expressed in bronchial epithelial cells after exposure to Pseudomonas aeruginosa (Mayer et.al. 2007 GEO dataset GSE6802). Of 34 genes documented to correct DF508 trafficking, 9 were directly linked by positive expression activation mechanisms to the inflammatory response. Looking at interactions among the results as a whole and in detail, it appears that an inflammatory response produces numerous changes which favor correct trafficking of DF508. Using knowledge management platforms to integrate results from multiple disciplines within the CF research community should accelerate our understanding of disease mechanism and identify therapeutic targets.