In silico identification of small molecules engaging the TNFR2-TNF-α interaction: a novel approach for targeting demyelinating diseases

Tumor necrosis factor alpha (TNF-) is a potent cytokine secreted by macrophages,. It is involved in immune and pro-inflammatory responses,, but under certain conditions it can also promote cellular proliferation and differentiation. TNF- exists in two isoforms, both arranged homotrimeric complexes: a a soluble form one,(solTNF), involved inthat participates to pathological mechanisms of demyelinating and neurodegenerative diseases preferably via TNF receptor 1 (TNFR1), and a transmembrane one form(tmTNF), which can mediate neurorepair and remyelination. TmTNF represents the constitutive and not processed form, while solTNF derives from the cleavage of the tmTNF extracellular domain by TNF- converting enzyme (TACE). Both isoforms interact with their receptors, TNF receptor 1 (TNFR1) and TNF receptor 2 (TNFR2), in their homotrimeric form through the receptor soluble domain. As previously described in literature, tmTNF preferably acts via TNFR2 signalling, while solTNF via TNFR1. Because Due to theof the protective role of TNFR2tmTNF in demyelinating central nervous systemdiseases (CNS), aim of this study is to identify , through an in silico approach, a set of small molecules which could act as ligandsselective TNFR2 ligands, able to selectively enhanceing TNFR2::tmTNF- engagement for and thus promoting its reparative effects. To achieve this goal, To achieve this goal, we firstwe first characterized in silico TNFR2 structure, underlying the differences with TNFR1. In particular, wWe found thate analyzed, in specific topological regions, the the interaction surface of both receptors, that are it is composed of four topological regions. We focused on regions three and four, which seem to be the most dissimilar regions between the two receptors and are characterized by opposite electrostatic potential surfaces,. Actually, in TNFR1 region three, the electrostatic potential surface is positively charged while in region four is negatively charged; conversely, in TNFR2, region three is negatively charged while region four is positively charged . These findings suggesting that the ability of the two TNF- isoforms to selectively engage either TNFR1 or TNFR2 could may partially depend on these electrostatic differences. To find molecules able to selectively recognize and engage TNFR2::TNF-complex, On this basis, wwe tested a large library of commercially available drug-like compounds against this complex, through performed aan in silico virtual high-throughput screening (HTS). Wanalysis against the TNFR2::TNF-complex, testing a large library of commercially available drug-like compounds specifically designed to influence protein::protein interactions (Asinex PPI database), to find molecules able to selectively recognize and engage this receptor. Among the approx. 11,000 tested molecules, we identified 20 compounds that in silico were able to bindinteract with high affinity with TNFR2::TNF- complex in the TNF- binding site but without affecting TNFR2::TNF- interaction. Then, to evaluate if these compounds were actually able to foster the TNFR2 engagement by TNF-, for each TNFR2::TNF-::ligand complex, we computed the difference between the G binding free energy value of the TNFR2::TNF- complex alone, and the G binding free energy of the complex in association with each investigated small moleculesligands (G), showing that. Results showed that aall the 20 compounds seem to enhance the affinity of TNF- for TNFR2., since the interaction energy of the complex in association with each single ligand is more negative than that of the complex alone. Finally, since these compounds should exhibit their activity in the central nervous system (CNS), to assess whether they could be efficiently delivered to the brain, their ability to target CNS was predicted in silico by computing three significant pharmacokinetic descriptors. Five out of the 20 selected compounds were characterized by a potential CNS activity., according to predicted in silico parameters such as of CNS, QPlogBB and QPPMDCK. To date, all the available approaches targeting the TNFR1/2::TNF- axis for promoting TNFR2 neuroprotection are based on the use of biotechnological moleculesdrugs. On this basis, Globally, our promising preliminary data suggest that it may be possible topave the way for the development of a new therapeutic strategy for demyelinating diseases that is based on TNFR2 engagement by small molecules with drug-like properties to promote its reparative effects.