Getting new drugs approved is a lengthy process and so, in the fight against COVID-19, our best hope is to use ones that have already been approved. Vittorio Limongelli and his group from the University of Lugano USI are using their world-leading computational methodologies to identify the best candidates for the job against a number of molecular targets.
Vittorio Limongelli’s group at the University of Lugano USI in Switzerland specialises in developing and applying computational methods to investigate how drugs interact with their molecular targets. Moving beyond standard molecular docking simulations, they have created a state-of-the-art methodology that reproduces the whole physical binding pathway of the drug to the molecular target.
Using rigorous free energy calculations, the team evaluates the different states that a drug can assume, allowing them to identify the drug binding mode as well as intermediate and transition states. Besides this thermodynamic information, the team’s methodology also characterises the kinetics of the drug-binding interaction. This provides information on the time that the drug spends in contact with the target, known as the residence time, which is fundamental to the efficacy of any drug.
“Like many other scientists, I felt an almost spiritual call this year to try and contribute our group’s knowledge towards helping reduce the suffering that we are currently seeing all over the world.”
Limongelli’s computational methodology is now well-established across the world, and was even published in Nature Protocols, a prestigious journal normally the preserve of experimental procedures. In addition to this, his group, together with a group led by Professor Siewert Marrink of the University of Groningen, has also developed the first-ever multi-scale model for drug binding to molecular targets. “Atomistic models suffer from many limitations in describing the reality of drug binding,” he explains. “Our model is able to include other aspects, such as drug concentration, that have previously been neglected.”
This multi-scale model represents an important advance in terms of bridging the gap between computational work and the real world it aims to describe. However, with great power comes great computational requirements. Resources like those provided by PRACE are fundamental to the methodology if results are needed in a short time period. With the global standstill brought about this year by the sudden rise of the COVID-19 pandemic, the world, now more than ever, needs answers fast.
“Like many other scientists, I felt an almost spiritual call this year to try and contribute our group’s knowledge towards helping reduce the suffering that we are currently seeing all over the world,” says Limongelli. “We have had to slow down or pause a number of our other research avenues in order to get to work on COVID-19, and it has required a lot of reading, but I believe our methodology is the perfect tool for finding treatments as quickly as possible.”
Schematic representation of the SARS-CoV-2 lifespan. The investigated targets are depicted in ribbons, showing their function in the virus replication mechanism (points A to E). The S protein is coloured in green (S), ACE2 in salmon (A), Mpro in skyblue (B), whereas RdRp (C) and RNA are in cadet blue and lime, respectively; the complex nsp10-nsp16 (D) is in turquoise and blue, respectively, whereas MAS1 is in orange (E).
Limongelli has been leading a drug-repositioning project supported by a PRACE allocation of
350 000 node hours on Piz Daint hosted by CSCS in Switzerland. Drug repositioning involves looking at drugs that have already been approved for safe use and seeing whether they have any therapeutic value for new targets. In this case, the group are analysing the effects of various drugs on some of the essential molecular players involved in the pathophysiology of SARS-CoV-2. Among them are the viral proteins that allow the virus to enter human cells and replicate there: main protease (Mpro), RNA-dependant RNA polymerase (RdRp) and 2’-O-methyltransferase (2’-O-MTase), and the human host proteins angiotensin-converting enzyme 2 (ACE2) and mitochondrial assembly 1 (MAS1).
This pool of targets was specifically chosen in order to cover the different phases of the disease, from the first contact the between the spike protein and the ACE2 receptor, to the MAS1 receptors involved in the more severe later stages of the disease in which hyper-inflammation occurs. Drugs that inhibit the interaction with ACE2 even have the potential to be used prophylactically, helping to decrease and contain viral infection in those who are at high risk of exposure.
Schematic representation of funnel-metadynamics applied to a sample case with the research workflow shown at the bottom. The computational technique is employed to select the best repurposed drugs in each investigated target.
The advantage of looking at already-approved drugs is that it greatly speeds up the process of getting to the stage where they can be used in clinical practice. “If you develop a new drug, it takes many years to even reach the clinical phase, and clearly we are in a situation where this is not helpful,” says Limongelli. “If we can identify drugs that are already on the market that are effective against SARS-CoV-2, we can skip the preliminary trials and go straight to the clinical phase.”
Combinations of drugs that hit different targets will be explored as part of the project to see if any synergies arise. Combining their own protocol with state-of-the-art machine learning models, the team will also explore whether any of the drugs are effective on more than one of the molecular targets being investigated. In many cases, drug repositioning does not identify particularly potent drugs, but if they exhibit even a weak activity against more than one important target – so-called multi-target drugs – they can be extremely effective.
The team has already identified a number of promising candidates. At the very top of their list is remdesivir, a drug originally developed to treat Ebola that was recently approved by the FDA in the US as the first authorised treatment for COVID-19. “Remdesivir appearing at the top of our list of candidates is encouraging as it means that our protocol has mirrored what has been found in the USA,” says Limongelli. “Our hope is that this means that the other candidates on our list are also likely to be effective in the fight against COVID-19.”
This article was also published in PRACE Digest 2020.
350 000 node hours on Piz Daint hosted by ETH Zurich/CSCS, Switzerland
1TB on the COVID-19 Computational Molecular Data Exchange solution hosted by ETH Zurich/CSCS, Switzerland