Characterising and classifying hypothalamus development

Understanding the development of the hypothalamus is important, due to its role as the central regulator of homeostasis. However, relative to development of other regions of the brain, characterisation and understanding of hypothalamus development is incomplete. Three important reasons for this are: (i) the hypothalamus is specified early compared to other brain regions, and then develops rapidly; (ii) the hypothalamus has a complex, anatomical structure even in the embryo; (iii) hypothalamic progenitor cells grow and migrate anisotropically. This non-linear growth makes it difficult to interpret downstream developmental events and molecular interactions that regulate early hypothalamus specification and regionalisation.

One promising way to investigate development of the hypothalamus is through combining computational methods and traditional embryological approaches. To this end, I begin this thesis by developing a method of fine-grained classification of the Hamburger Hamilton (HH) stage 10 chick embryo. I was able to train an accurate classifier despite a limited dataset, by testing a variety of biologically motivated data augmentation techniques. I encouraged confidence in the staging system and subsequent classifications by analysing and visualising the output of the classifier.

Using this classifier, I conducted a detailed morphological study of the developing hypothalamus at HH10 and surrounding stages, using both experimental embryology techniques and computational morphometric analyses. Using my increased understanding of the developing morphology, I characterised the expression of key hypothalamus morphogens: SHH, FGF10, and BMP2, as well as components of the SHH signalling pathway. I found that regionalisation between these morphogens occurred early and rapidly, with substantial heterogeneity in expression along both the anteroposterior and mediolateral axes.

Finally, I tested to what extent this regionalisation is neuroepithelium intrinsic using ex vivo culture. I found both anteroposterior and mediolateral regionalisation in culture, which suggests that these processes are self-organising in the neuroepithelium.

Overall, my thesis provides novel insights into early hypothalamic morphogenesis and molecular regionalisation, and shows through extension and use of the classifier how these complex processes may begin to be unpicked.