Computational approaches represent the future of rational drug design, and there has been no better time for this to be proven than during the onset of the pandemic. Professor Francesco Gervasio of University College London has been leading a many-pronged approach to finding therapeutics to fight the virus that has swept across the world, each of which relies on the power of HPC.
The COVID-19 pandemic has brought with it a great deal of uncertainty for people as they grapple with new ways of living and working. For many scientists, however, it has galvanised their sense of purpose, acting as a call to action to offer up their knowledge and expertise to support international efforts in combatting the disease.
Francesco Luigi Gervasio’s group at University College London develop computational algorithms that use free energy calculations to shine a light on the mechanistic details of complex biophysical events. When the UK was placed into lockdown earlier this year, Gervasio and his team decided to put their knowledge to good use and began discussions with potential collaborators in Europe and the USA. Realising that any investigations that they might wish to carry out would require large computational resources, the group applied to the fast-track call for COVID-19 proposals offered by PRACE. Gervasio is now leading two PRACE-supported projects exploring two different avenues for possible therapies.
The first of these projects, in collaboration with Prof. Shozeb Haider of UCL’s School of Pharmacy, Mount Sinai School of Medicine in New York and DE Shaw Research, among others, aims to design peptides that block the SARS-CoV-2 virus from entering the host human cells. The virus gains access to cells through the interaction of its spike protein with the human ACE2 receptor. By designing peptides that mimic the human receptor but bind even more potently to the viral spike protein, the team is creating a potential therapy to stop the virus at the very first hurdle. Creating such a peptide is easier said than done, however. Shorter peptides, while simpler, typically unfold and lose the necessary structure. Instead, the peptides have to be designed with helical structures so that they are sufficiently stable and do not unravel. The other challenge concerns redesigning the interface so that it binds more potently with the spike protein, as Gervasio explains: “We started off by extracting peptides from the human receptors, after which we added some hydrophobic residues to make the interface stronger and more stable. Our new design looked exceptionally stable in our simulations, but once we moved on to experimental validation we found a number of problems.”
As it turns out, the hydrophobic residues that were added to improve stability made the peptides difficult to produce in reality. “When we first tried to produce our peptides using E. coli, we found that their hydrophobic properties ended up killing the colonies,” he says. “Only after further laboratory work, which was logistically very difficult to arrange due to restrictions in the UK, were we eventually able to produce them. After carrying out circular dichroism spectroscopy, we were also able to show that the structure was stable.”
The second project led by Gervasio concerns one of the proteins on the SARS-CoV-2 virus known as nonstructural protein 1 (nsp1). Its role in the viral lifecycle is to bind and block production of specific proteins in the cell, which inhibits the immune response. From a drug discovery point of view, it represents a challenging target. “Nsp1 is essentially a sphere with a tail,” says Gervasio. “It binds to the Ribosome and blocks the synthesis of human proteins, including those that produced as a response to the infection. Problems arise when targeting this protein because the spherical shape means that there are no obvious cavities for drugs to bind to.”
Unfortunately for nsp1, another speciality of Gervasio’s group is the development of algorithms that look for hidden, or cryptic, cavities in such “undruggable” targets. “We put our cryptic site algorithms to work on nsp1 with the aim of finding small areas that drugs might be able to latch on to. Using special replica algorithms which encourage the opening of hidden cavities on molecules, we were able to identify a good potential site to target.”
The spike is a trimer that exists in two distinct conformational states resulting from the opening of the receptor-binding domain (RDB): ‘up’ or ‘down’. The ‘up ‘ conformation exposes the RBD, which is required for proteolytic processing and/or fusion of the S2 domain with the host cell membrane.
From a computational point of view, this project has also been a great success, and it is now down to experimental collaborators to move the research forwards. One group in London is using crystallography to validate the cavity, while a group in the USA is now designing potential compounds to bind to the site.
On top of these projects, Gervasio has been working on several other avenues of COVID-19 research. One of these is a drug repurposing project led by a clinical group working on cancer which, based on some pieces of evidence found in previous papers, has been exploring the possibility of using kinase inhibitors (normally used in the treatment of cancer) to block the interaction of the spike with the ACE2 receptor. “We have supported this group’s findings from cellular experiments with some solid models from docking simulations that show that these kinase inhibitors do indeed have the potential to block the spike protein,” says Gervasio.
The group are looking to publish the results of all of their work as openly as possible so that other groups can build upon what they have found. The research from both of the PRACE projects has the potential to help not only in the fight against COVID-19, but also other areas. “The peptides can be used in themselves as a therapy, but can also be conjugated with other materials like nanoparticles or other proteins to give more potent or entirely different effects.
“On the other hand, the nsp1 project, if it works, will represent an excellent example of using a rational strategy for targeting an “undruggable” target,” continues Gervasio. “Many of the crucial targets in complex diseases are similarly undruggable, so we might well have a strong proof of concept that will provide encouragement to the pharmaceutical industry that computational approaches, rather than large-scale experimental screening, should be spearheading the future of drug design.”