Safety is a key reason for failure of a drug. Often, drug failure occurs too late with approximately 90% failing in the clinic. Identifying these liabilities earlier will improve efficiency in the drug discovery and development process by decreasing the potential of late stage failure and reducing the escalating costs associated with developing new drugs. Despite this, translation of certain types of toxicity remain difficult to predict. For example, in the case of drug-induced liver injury (DILI), only 50% of human hepatic toxicities are picked up by preclinical animal studies. Therefore, more reliable in vitro human relevant models are required to identify this safety liability.
The first step in our understanding of drug-induced toxicity is linking the molecular initiating events to potential mechanisms by evaluating specific adverse outcome perturbations and relating these to organ toxicity. These molecular initiating fingerprints can be used for future hazard identification and risk assessment.
One area of toxicological research which is growing in interest is the field of transcriptomics. Transcriptomics is the study of mRNA molecules in the cell and it provides important information on how genes are expressed and interconnected. Recent technological breakthroughs have revolutionised transcriptomics. The introduction of RNA-seq (a method based on Next Generation Sequencing (NGS)) is one such advance. This powerful technique is able to provide a quantitative measurement of the entire transcriptome of the cell. Despite this, the real potential of the transcriptomics dataset can only be realised through sophisticated data processing techniques.
Cyprotex, in conjunction with parent company Evotec, has launched a new service which combines cell biology with next generation high throughput whole transcriptome sequencing using RNA-seq. By scaling up and industrialising this process, a more data-rich examination of the molecular phenotype of the cell can be achieved. Through our considerable expertise in bioinformatics, we have built a sophisticated data analysis platform known as PanHunter to manage the vast amounts of data produced and to map the sequencing data to the gene of interest. Machine learning and artificial intelligence are further used to interrogate the data and allow complicated questions to be easily resolved. By monitoring early cellular transcriptional response following exposure to a chemical, a sensitive in-depth view into early molecular initiating events with respect to efficacy or toxicity can be investigated.
Fully integrated in vitro transcriptomics prediction platform
Through our transcriptomics service clients can work with us on custom models, combined with transcriptomics and sophisticated data analysis tools to address specific needs.
The use of transcriptomics in DILI prediction has been recognised since the early 2000’s when surrogate gene expression markers were identified in human blood. Now the challenge is to detect DILI from human in vitro cell-based models. Our fully integrated platform is already showing promise in this respect. Primary human hepatocytes have been exposed to 128 reference compounds. Using AI modelling of transcriptional data, the accuracy of DILI prediction increased from 70% using established endpoints assays to 82% using the transcriptomics approach. Transcriptomics not only improves predictive accuracy but also enables a more detailed mechanistic insight. Our next steps involve expanding the database to include more compounds and different cellular models e.g., 3D microtissues, to further improve the prediction of DILI.
Using transcriptomics and artificial intelligence to improve DILI prediction
The role of transcriptomics is not just limited to DILI – it can be applied to other areas of toxicology such as cardiotoxicity and nephrotoxicity. Our transcriptomics service is flexible and can be customised for different organ-specific models and cell types. We can support all aspects of the transcriptomics workflow for an end to end in vitro toxicology transciptomics service. The workflow incorporates:
Through our unique highly automated, cost-effective processes, we can detect up to a million mRNA molecules per well and our generic protocol has been successfully applied to more than 25 cell types as well as 3D microtissue models.
High throughput transcriptomics produces vast quantities of complex data and specialised tools and software are required to manage, analyse and interpret the data. To address this, Evotec have developed the sophisticated PanHunter platform. It streamlines the entire process by storing and managing the sequencing data, performing quality control and differential expressions analysis, and evaluating the implications for downstream processes such as pathway regulation or gene network analysis. The PanHunter platform is able to combine data from various sources including, not only transcriptomics, but also genomics, proteomics, metabolomics and other specialised screens. Having access to this truly unique and powerful system transforms the workflow process - improving efficiency, reducing cost and ultimately enhancing the quality of the data through robust analysis and interpretation.