Modelling of gene regulatory and spatial mechanisms which pattern the early human neural tube

Abstract: Tissue patterning during development is thought to be driven largely by gene regulatory responses to spatial gradients of molecular signals, known as morphogens. In this thesis, early patterning of the human neural tube is studied using two different gene regulatory network models. The models portray patterning in the rostral-caudal direction as a response to gradients of WNT and FGF signalling. The first model is based on a tri-stable network topology which describes patterning into forebrain, midbrain, and hindbrain regions; the second model describes the formation and maintenance of the isthhmic organizer, and also subsequent midbrain formation. The two models were optimized against time-resolved bulk and single-cell RNA sequencing data from human embryonic stem cells undergoing 14 days of patterning in an in vitro microfluidics-based setup. In the setup, a GSK3 inhibitor is used to mimic the patterning effects of the in vivo early neural tube WNT signalling gradient. The resulting RNA data are in the form of time series over multiple GSK3 inhibitor concentrations. Clustering of the RNA data reveals that it is meaningful to group genes into the forebrain, midbrain, and hindbrain categories of the tri-stable model. It also reveals additional gene co-expression patterns within these categories, which is useful for further modelling work. The optimized network models are studied in reaction-diffusion simulations of WNT and FGF morphogen driven patterning in both a realistic 3D neural tube geometry, and in a 3D model of the tissue from the microfluidics setup. It is found that patterning of the tri-stable model is significantly influenced by WNT production, decay, and diffusion mechanisms, and that the WNT secretion locations of an earlier spatial 3D tube model, used as a foundation for this work, should be modified in order to produce a WNT gradient which is more similar to the in vivo WNT gradient shape. The isthmic organizer network model is found to accurately describe isthmic organizer formation at the intersection of regions with high OTX2 and GBX2 gene expression. Furthermore, the model isthmic organizer, once formed, maintains itself in the absence of an external WNT signalling gradient. WNT and FGF diffusion from the model isthmic organizer is found to induce midbrain identity adjacent to the organizer in approximately the right position. Overall, both models capture several key patterning features, and they will serve as useful stepping stones towards even more refined neural patterning models.