Periodic Reporting for period 2 - TOXIFATE (Future Toxicology: Better predicting Toxicant-induced cell fate.)
Periodo di rendicontazione: 2022-10-01 al 2024-09-30
Current global prosperity and modern health care rely upon the safe use of chemicals in industry, agriculture, and medicine. This prosperity is entirely dependent upon robust toxicology that identifies toxic chemicals so harm human health or the environment. The use of in vitro and computational approaches to predict chemical toxicity is revolutionizing toxicology right now. This revolution will have scientific, societal and ethical impact as it aims to improve toxicity assessment while reducing costs, increasing the number of chemicals that can be assessed and reducing animal use. A recent breakthrough was that computational approaches can sometimes outperform animal testing in predicting human toxicity, demonstrating the promise of this approach. However, our understanding of chemical-induced changes in cell and molecular biology is still rudimentary and we don’t have sufficient data to complete the revolution. The TOXIFATE project is contributing to the revolution by multi-disciplinary training in in vitro toxicology and computational approaches and generating a new and large dataset that will improve the prediction of muscle toxicity.
TOXIFATE is:
1) Generating new datasets that show the quantitative relationships between cell stress and cell death responses using new in vitro biosensor assays.
2) Providing multi-disciplinary training of EU toxicologists in innovative in vitro and in silico technologies.
3) Contributing to society by improving chemical safety by providing new ways to quickly, ethically, and economically identify chemical hazards.
Conclusions of the Action
The successful completion of a PhD in Cheminformatics and Toxicology, focused on myotoxicity, represents a significant scientific and societal achievement. This research aimed to address the challenge of predicting myotoxicity—a form of muscle damage caused by chemical or drug exposure—by developing computational models that leverage both phenotypic and transcriptomics data. The research was part of an intersectoral program that integrated academic, industrial, and research institution collaborations, resulting in an approach that is both scientifically robust and practically applicable.
The primary focus of the research was to develop predictive models capable of identifying myotoxicity—muscle damage caused by chemical or drug exposure—using phenotypic and transcriptomics data. Myotoxicity is a critical concern in drug development and toxicology, as it can result in severe side effects in patients, sometimes leading to the withdrawal of drugs from the market. The aim was to create robust models that could be used to predict myotoxic effects at early stages of development, thereby improving drug safety and reducing the likelihood of adverse outcomes in clinical settings.
TOXIFATE is:
1) Generating new datasets that show the quantitative relationships between cell stress and cell death responses using new in vitro biosensor assays.
2) Providing multi-disciplinary training of EU toxicologists in innovative in vitro and in silico technologies.
3) Contributing to society by improving chemical safety by providing new ways to quickly, ethically, and economically identify chemical hazards.
Conclusions of the Action
The successful completion of a PhD in Cheminformatics and Toxicology, focused on myotoxicity, represents a significant scientific and societal achievement. This research aimed to address the challenge of predicting myotoxicity—a form of muscle damage caused by chemical or drug exposure—by developing computational models that leverage both phenotypic and transcriptomics data. The research was part of an intersectoral program that integrated academic, industrial, and research institution collaborations, resulting in an approach that is both scientifically robust and practically applicable.
The primary focus of the research was to develop predictive models capable of identifying myotoxicity—muscle damage caused by chemical or drug exposure—using phenotypic and transcriptomics data. Myotoxicity is a critical concern in drug development and toxicology, as it can result in severe side effects in patients, sometimes leading to the withdrawal of drugs from the market. The aim was to create robust models that could be used to predict myotoxic effects at early stages of development, thereby improving drug safety and reducing the likelihood of adverse outcomes in clinical settings.
In the first half of the project TOXIFATE Doctoral Fellows (DFs) completed training at NUIG in the methodology of TEMPO-Seq transcriptomics and bioinformatics provided by Bioclavis (UK) and began large data analysis and cheminformatics training provided by ProtoQSAR (Spain). The DFs also received training in toxicology, including toxicity assays, advanced cell culture methods, automated high-throughput and high-content screening, and live cell imaging. The effects of thirty myotoxicants were assessed by phenotypic and transcriptomic profiling.
In the second part of the project these data these cell painting data were then used at ProtoQSAR to build computational models for myotoxicity prediction to:
1. Predict drug-induced cytotoxicity in muscle cells.
2. Compare and contrast XCiT networks and traditional machine learning approaches that rely on cell segmentation and feature extraction.
In addition, TOXIFATE generated the first proteomic analysis of proteolytic events during muscle cell death and differentiation identifying new proteins playing roles in muscle cell death and differentiation pathways.
This combined modelling approach enables the identification of specific biomarkers and pathways associated with myotoxicity, offering new insights into the mechanisms driving this adverse outcome. Such mechanistic understanding is crucial for developing safer drugs, as it allows for the identification of high-risk compounds early in the development process, reducing the likelihood of late-stage failures. Moreover, the models can be applied to existing chemical databases to screen for myotoxic potential in environmental chemicals, contributing to broader efforts in chemical safety and risk assessment.
In the second part of the project these data these cell painting data were then used at ProtoQSAR to build computational models for myotoxicity prediction to:
1. Predict drug-induced cytotoxicity in muscle cells.
2. Compare and contrast XCiT networks and traditional machine learning approaches that rely on cell segmentation and feature extraction.
In addition, TOXIFATE generated the first proteomic analysis of proteolytic events during muscle cell death and differentiation identifying new proteins playing roles in muscle cell death and differentiation pathways.
This combined modelling approach enables the identification of specific biomarkers and pathways associated with myotoxicity, offering new insights into the mechanisms driving this adverse outcome. Such mechanistic understanding is crucial for developing safer drugs, as it allows for the identification of high-risk compounds early in the development process, reducing the likelihood of late-stage failures. Moreover, the models can be applied to existing chemical databases to screen for myotoxic potential in environmental chemicals, contributing to broader efforts in chemical safety and risk assessment.
Advance over the state-of-the-art
The program’s intersectoral nature was critical to its success. The project was conducted in close partnership with an industrial collaborator, who hosted the doctoral fellows and provided access to proprietary data and practical insights. This industry exposure enabled the development of models that not only demonstrated high predictive performance but were also aligned with the requirements and constraints of real-world applications. The involvement of the industrial partner ensured that the research was relevant to pharmaceutical safety evaluation, where early identification of myotoxic compounds is crucial for preventing adverse drug reactions in clinical trials and ensuring patient safety. In addition to the industrial collaboration, the program included specialized training from Bioclavis and the Fraunhofer Institute for Translational Medicine. Bioclavis, with its expertise in transcriptomics, provided hands-on training in analyzing gene expression data, allowing the researchers to identify molecular signatures associated with myotoxicity. This capability enhanced the predictive models by enabling the incorporation of transcriptomics data, providing early indicators of myotoxic risk. The training received at the Fraunhofer Institute further broadened the scope of the research by offering insights into the screening of molecular libraries. This expertise helped the researchers understand the interactions between different compounds and biological targets, thereby strengthening the mechanistic basis of the developed models.
The scientific impact of this research is substantial. By combining phenotypic and transcriptomics data, the models developed can predict myotoxicity with a high degree of accuracy and interpretability. This integrative approach provides a more comprehensive understanding of the biological mechanisms underlying myotoxicity, offering valuable insights into how different chemicals or drugs might cause muscle damage. These findings contribute to the development of safer drugs and support regulatory agencies in evaluating the safety of environmental and industrial chemicals.
Societally, this research holds promise for reducing the incidence of adverse drug reactions and improving public health and safety. The collaborative nature of the program, bringing together academic research, industrial expertise, and specialized training, exemplifies the benefits of intersectoral partnerships in tackling complex scientific problems and translating research into real-world applications.
The program’s intersectoral nature was critical to its success. The project was conducted in close partnership with an industrial collaborator, who hosted the doctoral fellows and provided access to proprietary data and practical insights. This industry exposure enabled the development of models that not only demonstrated high predictive performance but were also aligned with the requirements and constraints of real-world applications. The involvement of the industrial partner ensured that the research was relevant to pharmaceutical safety evaluation, where early identification of myotoxic compounds is crucial for preventing adverse drug reactions in clinical trials and ensuring patient safety. In addition to the industrial collaboration, the program included specialized training from Bioclavis and the Fraunhofer Institute for Translational Medicine. Bioclavis, with its expertise in transcriptomics, provided hands-on training in analyzing gene expression data, allowing the researchers to identify molecular signatures associated with myotoxicity. This capability enhanced the predictive models by enabling the incorporation of transcriptomics data, providing early indicators of myotoxic risk. The training received at the Fraunhofer Institute further broadened the scope of the research by offering insights into the screening of molecular libraries. This expertise helped the researchers understand the interactions between different compounds and biological targets, thereby strengthening the mechanistic basis of the developed models.
The scientific impact of this research is substantial. By combining phenotypic and transcriptomics data, the models developed can predict myotoxicity with a high degree of accuracy and interpretability. This integrative approach provides a more comprehensive understanding of the biological mechanisms underlying myotoxicity, offering valuable insights into how different chemicals or drugs might cause muscle damage. These findings contribute to the development of safer drugs and support regulatory agencies in evaluating the safety of environmental and industrial chemicals.
Societally, this research holds promise for reducing the incidence of adverse drug reactions and improving public health and safety. The collaborative nature of the program, bringing together academic research, industrial expertise, and specialized training, exemplifies the benefits of intersectoral partnerships in tackling complex scientific problems and translating research into real-world applications.