From model to medicine: Personalized approaches to inborn errors of metabolism

Abstract

Our bodies run on thousands of chemical reactions that convert food into energy and building blocks. Enzymes drive these reactions. In patients with an inborn error of metabolism (IEM), DNA mutations lead to enzyme malfunction, causing accumulation of toxic substances or shortage of essential molecules. More than a thousand different IEMs are known, with roughly 800 affected children born in the Netherlands each year. Diagnostics are improving rapidly, but for many IEMs effective treatment is still lacking. This thesis sought to narrow that gap, using patient-derived cell models to study disease mechanisms and test therapies in the laboratory. Part 1: Protein synthesis disorders (ARS deficiencies) Proteins are assembled from amino acids in a precise order, driven by aminoacyl-tRNA synthetase (ARS) enzymes. A deficiency of any one ARS enzyme causes severe disease. Studying >100 patients revealed recurring patterns: growth delay, neurological problems, and liver disease, that worsen when protein demand is high or enzyme activity drops during infections (chapter 2). Supplementing the diet with the specific amino acid handled by the deficient enzyme produced clear clinical improvement in four patients (chapter 3). More fundamentally, we discovered that cells under amino acid shortage deliberately incorporate a "wrong" amino acid to keep protein production going: speed over accuracy. We propose this might be a general cellular survival strategy (chapters 4 and 5). Part 2: Liver models for metabolic research Many IEMs primarily affect the liver, but culturing patient liver cells long-term is currently infeasible. Bile-duct-derived organoids (ICOs) are an accessible alternative, though their utility depends on whether the relevant metabolic pathway is sufficiently active in these cells (chapter 6). To help researchers select the right patient-derived model, we built HLCompR, a benchmarking tool that systematically compares liver-like cell models to primary hepatocytes for any function of interest (chapter 7). We then developed HeLLOs, an improved liver organoid model that outperformed existing models in key hepatocyte functions including bile acid synthesis and drug toxicity testing (chapter 8). Finally, we showed that the energy metabolism of liver progenitor cells actively governs their replication and identity: pyruvate oxidation proved essential for cell proliferation, but also for differentiation towards hepatocytes (chapter 9). Part 3: Precision genome editing When no treatment exists, correcting the genetic mutation in a patient's own cells offers a future route. Using prime editing, a "find-and-replace" approach to DNA editing, we corrected disease-causing mutations in patient-derived organoids for multiple IEMs, including DGAT1 deficiency, cystic fibrosis, and copper storage disease and achieved functional rescue. Whole-genome sequencing confirmed no off-target changes, a prerequisite for clinical use (chapter 10). We also developed fluoPEER, a mutation-specific fluorescent reporter system that enables rapid optimization of prime editing designs for any mutation (chapter 11). Taken together, this thesis shows that patient-derived models combined with precision gene editing bring personalized therapy for IEM patients closer, but scaling to the full breadth of IEM mutations remains the central challenge ahead.