Computational screenings to analyze the 'pain protein'

Computational chemistry and artificial intelligence may save time and resources in identifying the causes of nervous system disorders. This is proved by a study, just published in the prestigious Scientific Reports journal, proposing computational 'screening' to identify which genetic mutations of a specific protein could cause the disease.

The protein in question is called NaV1.7 and is known to neuroscientists as one of the fundamental communication channels between the peripheral and central systems: it regulates the 'volume' of pain that reaches the brain. Some mutations can make people unable to feel pain, putting them at risk. Other mutations can lower the threshold and cause unwarranted pain.

An international and multidisciplinary team, coordinated by computer scientist Marta Simeoni and physicist Achille Giacometti from Ca' Foscari University of Venice, has developed and tested for the first time a method of predicting which mutations deserve diagnostic investigations, distinguishing them from genetic variants of the protein that do not alter its functioning. This is the first step towards an advanced method that will enable the targeted use of electrophysiological analysis, which requires a lot of resources and many months of work, focusing on the most worthy cases.

To do so, the researchers applied computational chemistry and machine learning to a series of 85 NaV1.7 mutations provided by the Carlo Besta Neurological Institute of Milan. These techniques recognized structural patterns associated with the onset of painful neuropathies, thus anticipating the possibility of developing increasingly specific drugs with fewer side effects.

Among the major obstacles in developing drugs, there is the current inability to produce specific drugs for this protein: eight other proteins homologous to Nav1.7 have been discovered in the human body. All of them control the diffusion of sodium ions across the membrane in response to changes in potential, but are encoded by nine different genes within our DNA.

Being homologous to each other, they also share a very high structural similarity, and this makes it extremely difficult to develop drugs that are specific for only one of these proteins without altering the functionality of the other homologous proteins that are working properly.

"We think that the computational pipeline we have adopted can become a methodology applicable even to other proteins," Marta Simeoni explains, "but this needs to be verified and therefore there is still a lot of work to be done."

“We intend to continue in this direction by implementing a veritable Digital Health Lab," adds Achille Giacometti, who is director of the international European Centre for Living Technology (ECLT) in Venice.

Two Ca' Foscari students also collaborated in the research: Alberto Toffano and Giacomo Chiarot got the chance to participate in the study alongside scientists from the ECLT, the Besta Institute, the University of Maastricht (Holland) and Yale University (United States).

The image in this article was created from a model of the protein thanks to a tool developed by Ca' Foscari final year student Sara Corazza and available for all devices.