METABOLIC MODELLING BASED APPROACH TO IDENTIFY TAILORED METABOLIC DRUGS

Doctoral thesis (2020)

Cancer is one of the leading causes of death worldwide with no efficient cure. Even though the currently existing cancer therapies show promising results in early detected cancers, the treatment of late ... [more ▼]

Cancer is one of the leading causes of death worldwide with no efficient cure. Even though the currently existing cancer therapies show promising results in early detected cancers, the treatment of late state cancers and metastases still presents many obstacles. The development of treatment resistance and non-responsive cancer hampers the already time-consuming endeavour of drug development. To overcome these hurdles, researchers started to turn to computational approaches to unravel the mechanisms of cancer in terms of onset, development, resistance mechanism, metastasis, and immune invasion among others. The advent of systems biology came not only with the ability to generate large amounts of data but also novel approaches and techniques to analyse them. For example, metabolic modelling approaches can integrate -omics data into a computational representation of metabolism and reconstruct context-specific metabolic models via model-building algorithm such as FASTCORE (Vlassis et al., 2014), FASTCORMICS (Pacheco et al., 2015), and rFASTCORMICS (Pacheco et al., 2019b). Since then, context-specific metabolic models have gained a large area of application that ranges from understanding basic metabolism of microbes, in silico engineering and optimisation of different bacterial strains to applications in human diseases such as biomarker identification and drug target prediction in several diseases, including cancer. In this thesis, I give an overview on metabolic modelling approaches with a focus on cancer and different methods of drug discovery based on the prediction of in silico targets that enable finding cancer-specific targets to advance personalized medicine. To this end, I have developed a drug prediction workflow that has been successfully used to predict and validate three drugs for repurposing in colorectal cancer using context-specific metabolic models reconstructed via rFASTCORMICS (Pacheco et al., 2019b). Furthermore, we have reconstructed 10 005 context specific models for cancer and its controls to investigate the metabolic differences and rewiring strategies used in cancer. A second application of the workflow for melanoma is currently in progress, several drugs have been predicted for repurposing in melanoma and will be tested in vitro on different melanoma cell lines. Additionally, the most promising drugs will also be tested in combination with current melanoma treatment. [less ▲]

Towards the Integration of Metabolic Network Modelling and Machine Learning for the Routine Analysis of High-Throughput Patient Data

in Automated Reasoning for Systems Biology and Medicine (2019)

The decreasing cost of high-throughput technologies allows to consider their use in healthcare and medicine. To prepare for this upcoming revolution, the community is assembling large disease-dedicated ... [more ▼]

The decreasing cost of high-throughput technologies allows to consider their use in healthcare and medicine. To prepare for this upcoming revolution, the community is assembling large disease-dedicated datasets such as TCGA or METABRIC. These datasets will serve as references to compare new patient samples to in order to assign them to a predefined category (i.e. ‘patients associated with poor prognosis’). Some problems affecting the downstream analysis remain to be solved, the bottleneck is no longer data generation itself but the integration of the existing datasets with the present knowledge. Constraint-based modelling, that only requires the setting of a few parameters, became popular for the integration of high-throughput data in a metabolic context. Notably, context-specific building algorithms that extract a subnetwork from a reference network are largely used to study metabolic changes in various diseases. Reference networks are composed of canonical pathways while extracted subnetworks include only active pathways in the context of interest based on high-throughput data. Even though these algorithms can be part of automated pipelines, to be applied by clinicians, the model-building pipelines must be coupled to a standardized semi-automated analysis workflow based on machine learning approaches to avoid bias and reduce the cost of diagnostics. [less ▲]

Integrated In Vitro and In Silico Modeling Delineates the Molecular Effects of a Synbiotic Regimen on Colorectal-Cancer-Derived Cells

in Cell Reports (2019), 27

By modulating the human gut microbiome, prebiotics and probiotics (combinations of which are called synbiotics) may be used to treat diseases such as colorectal cancer (CRC). Methodological limitations ... [more ▼]

By modulating the human gut microbiome, prebiotics and probiotics (combinations of which are called synbiotics) may be used to treat diseases such as colorectal cancer (CRC). Methodological limitations have prevented determining the potential combina- torial mechanisms of action of such regimens. We expanded our HuMiX gut-on-a-chip model to co-culture CRC-derived epithelial cells with a model probiotic under a simulated prebiotic regimen, and we integrated the multi-omic results with in silico metabolic modeling. In contrast to individual prebi- otic or probiotic treatments, the synbiotic regimen caused downregulation of genes involved in procarci- nogenic pathways and drug resistance, and reduced levels of the oncometabolite lactate. Distinct ratios of organic and short-chain fatty acids were produced during the simulated regimens. Treatment of primary CRC-derived cells with a molecular cocktail reflecting the synbiotic regimen attenuated self-renewal ca- pacity. Our integrated approach demonstrates the potential of modeling for rationally formulating synbi- otics-based treatments in the future. [less ▲]

Prediction of drug targets using metabolic modelling

Bachelor/master dissertation (2016)

Cancer, as one of the leading causes of death worldwide, is a disease characterized by the abnormal and uncontrolled proliferation of cells. Currently available anti-cancer drugs come with a variety of ... [more ▼]

Cancer, as one of the leading causes of death worldwide, is a disease characterized by the abnormal and uncontrolled proliferation of cells. Currently available anti-cancer drugs come with a variety of different side effects reducing the quality of life of cancer patients. Due to these severe side effects in anti-cancer therapy it is important to find a compromise between killing the cancer cells (efficiency) and not affecting the healthy cells (toxicity) to improve the quality of life of those patients. There exist different methods of finding new drug targets in cancer such as the in vitro development of new drugs which is very time consuming and expensive. The in silico prediction of targets, on the other hand, is fast and cost effective and allows to make a pre-selection of drug targets based on candidate genes. In this work, I propose a new workflow which implements metabolic modelling for finding metabolic drug targets in cancer. Therefore, context-specific models for cancer (including primary and metastatic melanoma) and healthy controls were reconstructed from Recon 2 (a genome scale metabolic model) using FASTCORMICS and two different expression datasets. In silico single gene deletion was performed in the models to search for potential candidate genes which are essential in cancer (reduce biomass production by 50%) but not in healthy (do not affect ATP production). In a second step, (approved) drugs targeting metabolic genes and their side effects, were extracted from the DrugBank, STITCH and SIDER through data mining and mapped to the metabolic network. A total of 65 possible drug targets have been found. These targets include genes which are known targets for chemotherapeutic agents such as the thymidylate synthase (TYMS), the fatty acid synthase (FASN) or dihydrofolate reductase (DHFR). Furthermore, two anti-cancer agents have been predicted for FASN which have already been proposed for the treatment of cancer. [less ▲]