Metalloproteins investigation for the development of new inhibitors

Nowadays, drug-resistant bacteria are becoming a real threat to our society. The chance of dangerous pathogens developing resistance to classical antibiotics and the risk of being unable to fight this impending menace without innovative drugs is drawing ever closer. Our study aims to develop a computational protocol to speed up the drug discovery pipeline for a new class of antibiotics, capable of inhibiting the specific metalloproteins crucial for bacteria survival. Every year drug-resistant tuberculosis causes the deaths of hundreds of thousands of people worldwide. The main targets of our new generation of antitubercular drugs are the β Carbonic Anhydrases, which are selectively expressed by the pathogen. To accomplish this drug discovery task, we started from scratch to build completely new chemical entities. The plumb of the available chemical space, which is esteemed to contain around 1023 drug like compounds, for the “de novo” discovery of a new candidate is a chimerical challenge, so efficient methods are needed to explore it. Deep learning algorithms, like Generative Adversarial Networks, can speed up this process, enabling the opportunity to accelerate the search of the chemical space. This was how we intended to take advantage of AI, to gain access to a reasonable number of promising starting points for our drug design purpose. Then, we performed a comparative virtual screening procedure, that involved different Docking suites that explicitly consider metal-ligand interactions, to assess the binding affinity of the generated and optimized hit compounds. Lastly, we managed to set up an alchemical free energy calculation protocol to obtain more accurate free energy predictions of the metal-binding ligands, to guide the hit-to-lead optimization. Regarding the theoretical framework necessary to simulate the nuanced dynamic behaviour of the metalloproteins target, we are involved in the setup of tailored Force Fields and QM/MM MD simulations, to analyze the coordination bond formation mechanisms and the energy. Indeed, we managed to design a low nanomolar inhibitor, selective for the pathogenβ2 Carbonic Anhydrase, whose activity is going to be tested in vivo.