Reconstruction of regulatory networks to predict gene function in pancreatic development and disease

The complex gene regulatory networks (GRNs) establishing acinar, ductal, and endocrine lineages during pancreatic organogenesis are poorly understood. Albeit each compartment gives rise to a different set of diseases, there is a genetic overlap between pancreatic development and both pancreatic cancer and diabetes. Therefore, a better understanding of transitions during normal development and disease could inform cell engineering efforts and lead to improved therapies. By reconstructing pancreatic GRNs, we aim to understand the perturbed programs in monogenic diabetes and boost cell engineering.

We applied Boolean network models which are widely used models for regulatory processes. Initial modeling of the regulatory functions was based on manual and text-mining approaches. We used tools such as BoolNet and ViSiBooL to model and simulate the Boolean networks. Next, we focused on real-valued time-series of RNA-seq data over the different pancreas developmental stages. After binarization (BiTrinA), the regulatory dependencies between components are inferred and integrated into the network.

First, we developed an initial Boolean network model of HNF6 signaling in pancreatic progenitor cells. This Boolean network model is based on extensive literature research and comprises 38 key genes. Then, we refined the initial Boolean network using the previously described time-series omics data. Our extended network model was validated using knowledge from interaction databases and proved overall accurate. Finally, we apply this model to predict developmental changes during pancreas differentiation in cells depleted of specific genes. We plan to simulate loss of HNF6, a novel diabetes gene recently identified and well characterized by our group, in order to predict transcriptional alterations. In the future, we will further expand this network with the input of bulk and single-cell-resolved transcriptional and chromatin dynamics data to elaborate the pancreatic lineages.

Our Boolean network reconstructing several stages of pancreas development is a valuable tool to predict gene function in developing and diseased pancreas. This model provides the theoretical basis to understand the organization of the human pancreas - the principle to decode pancreatic disease.

Publication History

Article published online:
08 September 2020