Multi-layered signaling networks in uncovering hidden interactions and key regulators in cancer, inflammatory bowel disease and during infection

Networks describe the relationships of the elements of biological systems using edges and nodes. However, the resulting representation of the system can sometimes be too simplistic to usefully model reality and provide new mechanistic insight. By combining several different interaction types within one larger multi-layered biological network, we developed resources such as SignaLink, OmniPath, NRF2ome, Autophagy Regulatory Network and SalmoNet to provide a more nuanced view than those relying on single-layer networks. Multi-layered networks display connections between multiple networks (i.e.: protein-protein interactions and their transcriptional and post-transcriptional regulators), each one of them describing a specific set of connections. We used these resources in the last couple of years to investigate cancer signaling networks, and recently inflammatory bowel disease and host-microbe interactions in the gut. Using integrated networks, we proposed new therapeutic approaches to improve success rates and to identify suitable proteins as novel, alternative drug targets for solid tumors. We designed a computational approach, combining mutation and differential expression data with network information, to analyse the interactions of cancer-related proteins in colon, breast, liver and lung cancer. We found that first (direct) neighbours, not linked previously to the given cancer type, are similarly important as mutated proteins known to be involved in cancer development. We also found 223 drugs already in the clinic targeting these proteins but not yet used against cancer as their oncology relevance was hidden so far. In a similar study, we investigated ulcerative colitis, a type of inflammatory bowel disease. This time we did a precision medicine approach, using patient-specific SNP-profiles for integration with multi-layered networks. We demonstrated that while most of the SNP affected proteins are not related to ulcerative colitis associated functions, their interactor partners (not mutated in patients) are annotated to central pathways with biomedically relevant regulatory roles. Moreover, we were able to cluster ulcerative colitis patients having nearly the same symptoms but possessing cluster-specific mutation sets effecting different pathways and cell types. Finally, we also combined these multi-layered network resources to investigate host-microbe interactions. We predicted and prioritised a list of potential connections between the host autophagy machinery and a gut pathogen, Salmonella. These connections could serve as mechanisms how Salmonella is modulating the host upon infection. We established an intestinal organoid system to test and validate these predicted interactions using multi-omics readouts, including small RNA profiling and single cell sequencing.