ETUDE DE LA COEVOLUTION DE PROTEINES POUR L’IDENTIFICATION DE REGIONS D’INTERACTIONS IMPLIQUEES DANS L’ACTIVATION DU FACTEUR DE CROISSANCE TGF-β

Background

Chronic liver diseases (CLD) are diseases of long duration and slow progression and represent a major public health issue, accounting for approximately 2 million deaths per year worldwide. Liver fibrosis is the final outcome of CLD and has been mainly associated with hepatitis B and C viral infection, alcohol drinking and non-alcoholic fatty liver disease including non-alcoholic steatohepatitis. Complications of cirrhosis, the end-stage of fibrosis, include ascites, variceal bleeding, encephalopathy, and liver cancer and is the fourth most frequent cause of death in adults in central Europe (Roehlen et al, 2020).

The Transforming Growth Factor β protein (TGF-β) is widely regarded as the major pro-fibrogenic agent. While TGF-β regulates cellular homeostasis in normal tissue, its activity is modified in chronic liver diseases. It triggers the excessive deposit of extracellular matrix (fibrosis) leading to the development of hepatic insufficiency. Targeting the deleterious effects of TGF-β without affecting its physiological role is the common goal of therapeutic strategies. However, the identification of specific targets in the extracellular matrix (ECM) remains challenging because of the pleiotropic effects of TGF-β linked to the complex nature of its extracellular activation and signaling pathways.

Numerous modeling approaches have been developed to describe the behavior of TGF-β-associated signaling pathways (Théret et al, 2020a review). However, not all these models consider the extracellular components that regulate TGF-β bioavailability and constitute more accessible therapeutic targets (Robertson et al, 2016). In this context, we have recently integrated knowledge from the literature (over 100 publications) describing the molecular interactions involved in TGF-β activation. This protein-protein interaction (PPI) network contains 30 molecules and around 300 interaction rules (Théret et al, 2020b). In some cases, the precise interaction sites are known and can be experimentally investigated for targeting with therapeutic compounds. However, in most other cases the precise regions for the partners’ interaction are lacking and some of the reported PPIs can be false positives caused by indirect interactions and thus without such regions. 

Objectives

We propose in this thesis to use and improve recent phylogenomic and coevolutionary approaches to predict the interacting regions and to better characterize and understand the PPIs involved in the TGF-β activation.

Approaches

We previously addressed the problem of dealing with the typical paralogous expansions and the multidomain architectures of the ECM proteins by applying a phylogenomic approach (Dennler et al, 2023). We have predicted conserved sequence signatures resulting from selective pressures maintaining key phenotypes from a given clade ancestor. Because major processes in the ECM involve interactions between proteins, we considered PPIs as the key phenotypes and associated them with conserved sequence signatures in the ADAMTS/TSL protein family (Dennler et al, 2023). The method predicts the sequence signature putatively involved in the interaction in one of the two interacting partners. The method was applied to the 5 ADAMTS/TSL paralogs involved in the TGF-β interaction network, and to the fibulin family whose members are also involved in this network (Master II, Elisa Chenel). It will be applied to the other TGF-β interacting proteins to investigate their respective conserved regions

Moreover, it is recognized that interacting proteins are coevolving (Szurmant et al 2018), meaning that the Darwinian selection for maintaining the interaction during evolution applies simultaneously on the sites from both the interacting proteins that are involved in their interaction. Detected coevolution patterns affecting the TGF-β interacting proteins could serve as a proxy to refine and infer new interacting regions. This task could benefit from recent advances in the development of coevolution-based approaches in Bioinformatics. One can cite in particular, the phylogenetic profiling methods which are based on the hypothesis that functionally related genes are associated with similar evolutionary pressures. Such methods were recently used with a machine-learning approach (Stupp D et al, 2021) to identify pairs of coevolving proteins. The evolution of the PPI networks can also be modeled through different mechanisms: sequence evolution and coevolution for interacting proteins, gene gain by gene duplication, de novo gene birth, or horizontal transfer, gene loss, and interaction gain or loss (Ghadie et al, 2018). At the level of residues, the docking server InterEvDock3 allows to predict the structure of protein-protein interactions using evolutionary information (Quignot et al, 2021) while the BIS2Analyzer is dedicated to a phylogeny-driven coevolution analysis of protein families with different evolutionary pressures (Oterii et al, 2022).

We propose to investigate how current methods for detecting coevolution patterns can be used or adapted to identify interaction regions within the TGF-β activation network, given the many paralogs and complex domain compositions of its interacting proteins. More specifically, the binding sites that we have already identified are expected to be detected as coevolving, providing a control on the predictions. The interacting proteins without binding sites yet identified will be investigated for their coevolving regions, considered as putative interacting sites. We expect the new interactions predicted on the basis of their coevolution to provide deeper insights into the interaction network of the TGF-β, a key actor of liver fibrosis and tumor progression. We also expect the developed methods to be useful for studying the other PPI networks of interest in the fields of biology and health.