State-dependent control of cell-fate transition

Thesis

Cell-fate transitions underpin development, regeneration and disease, and are conventionally explained by identifying molecular drivers that induce new cellular identities. Although this driver-centric framework has identified many regulators of cell fate, it does not explain why cells exposed to comparable inducing signals frequently produce different outcomes. This thesis tests the hypothesis that cellfate transitions are governed not by induction alone, but by state-dependent moderation in which the effect of an inducer depends on the current cellular state. 

To address this, I developed SIGMAVI (Selective Interaction Generative Mapping Adaptive Variational Inference), a computational framework that identifies resistance programmes from snapshot single-cell transcriptomic data and estimates how they modify the relationship between induction and fate transition. 

In embryonic epicardial epithelial-to-mesenchymal transition (EMT), SIGMAVI identified distinct inducer and resistance programmes that jointly predict fate transition, and identified WT1 as a principal regulator of resistance through maintenance of chromatin accessibility at resistance-associated regulatory elements. Developmental analyses suggest that declining epicardial competence arises through progressive depletion of permissive cellular states rather than uniform acquisition of resistance. The same regulatory architecture was recovered following adult cardiac injury, suggesting that regenerative failure may not occur at EMT initiation. 

The framework subsequently generalised beyond epicardial development. During induced pluripotent stem cell (iPSC) reprogramming, resistance distinguished converting from non-converting cells independently of experimental time and transition position, revealing structured variation underlying heterogeneous reprogramming outcomes. Across four cardiomyopathies of distinct aetiology, the same moderation relationship was independently recovered despite largely disease-specific inducer and resistance programmes.

Together, these findings support a model in which the effectiveness of induction is conditional on the state of the responding cell. The conserved feature across systems is therefore not a particular molecular programme, but a decision architecture linking induction, resistance and fate. This provides a framework for understanding heterogeneous cellular responses and shifts the study of cell fate from asking only what drives a transition towards what determines whether that drive succeeds.

Authors: Dang, QM

• DOI: 10.5287/ora-8gz9zjpqg
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: development and regeneration; cell-fate transition; single-cell genomics; variational inference; stem cell biology; cardiac disease
• Subjects: Machine learning; Regenerative medicine; Medical sciences; Development; Computational biology