The in silico limb : data-based spatio-temporal modelling of vertebrate limb organogenesis

With the advantages of characterized signaling centers, known regulatory networks and characterized mutants, vertebrate limb bud is a paradigm to study organogenesis. Classical and molecular experiments have identified major key components and have provided snap shot information about the underlying regulatory processes. The mechanisms that control the size and shape of the limb bud have to some extent remained elusive and quantitative data is missing. To that end, I have acquired high quality quantitative data from defined stages of mouse and chicken limb bud development. In addition, my research aimed to gain an integrative understanding of the limb bud development utilizing an in silico modeling approach. This was done by using real geometries and the gene expression domains of particular genes.
For generating a quantitative dataset of mouse and chicken forelimb and wing bud growth, I have combined limb bud specific reporter expression, FACS analysis and 3-D imaging. We find that the growth and proliferation rates decline over time in a way that a biphasic growth behavior is observed. After the initial expansion, a distinct second phase starts around the stages when the hand plates are formed. Even though Sox9 positive cells have a lower proliferation rate, we show that the increasing conversion of Sox9 negative cells into Sox9 positive chondrogenic progenitors alone cannot be responsible for the observed lowering in growth rates. This is due to the fact that the fraction of Sox9 positive cells remains constant at the time when the growth rate drops and the proliferation rates decrease in both Sox9 positive and negative populations. I propose that the decrease of growth and proliferation rates over time is independent of the known limb patterning mechanism.
For generation of the in silico limb bud model simulations, I first collected a 4-D gene expression data set in mouse limb buds by combining RNA in situ hybridization with Optical Projection Tomography (OPT) analysis between embryonic day 9 (E9) and E12.5. Using these high-resolution image data sets, I extracted limb bud geometries to allow solving reaction-diffusion equations on these scaffolds. Our in silico model simulations show the spatio-temporal kinetics of the molecular interactions that control initiation, propagation and termination of mouse limb bud development. Moreover, our model simulations are able to explain behaviors of various mouse limb bud mutants, which alter the levels and spatio-temporal kinetics of gene expression.
Based on predictions from the OPT analysis and the in silico simulations, I have also evidence of an additional pathway that appears to regulate Gremlin1 expression during limb bud development. Firstly, using Prx-Cre mediated mesenchymal inactivation of the WNT pathway effector β-Catenin and specific inhibitors of the WNT pathway, I have shown that WNT pathway activity is required for the characteristic distal-anterior expansion of Grem1 expression during progression of limb bud development E10.75 onwards. Using Hoxa13-Cre inactivation of the WNT pathway inactivation, I have shown that WNT signalling is also needed for characteristic Gremlin1 expression during hand plate development.