Simulations of Coronavirus Entering Cells Could Boost Drug Design
And… halt! When it comes to stopping a virus it its tracks, it’s crucial to know exactly how and where the tiny intruder actually gets into an organism. For the first time, researchers have generated a complete molecular model of the specific elements of coronavirus that penetrate our cells’ membrane — and understood how and why the penetration happens.
Understanding the process of the penetration and fusion of the virus’s and the host’s membranes at the atomic level is vital to the design of molecular inhibitors that would prevent the virus from entering the cells.
Early in the COVID-19 outbreak scientists determined that, like other coronaviruses, SARS-CoV-2 contains the now-infamous spike protein it uses to latch onto its victim’s cell. When the spike has attached itself to a protein at the cell’s surface and is ready to penetrate, it takes on a “fusion-ready state.” It does so by organizing its fusion peptides — chains of amino acids that form proteins — into a spearhead that performs the actual insertion into the membrane.
But what this spearhead looked like and how and where exactly it engaged the membrane wasn’t clear — until now.
A team from Cornell University’s Weill Cornell College of Medicine in New York City has used the IBM-developed supercomputer AiMOS at Rensselaer Polytechnic Institute (RPI), via the COVID-19 High Performance Computing Consortium, to identify the fusion peptide’s shape and preferred mode of attack when it penetrates the membrane. To discover this, the researchers carried out massive computations of the structural dynamics of fusion peptide, and revealed how, where, and why the peptide enters the cell.
The discovery explained previous findings on coronaviruses that showed that calcium ions — atoms that have become positively charged and that usually surround the cell — help infectivity. “The fusion peptide contains many clues to possible ways in which it could bind calcium, but without knowing the structural details of the fusion peptide as it prepares to pierce the membrane it was not known how and why this would help,” says Harel Weinstein, a biophysicist at Weill Cornell who leads the study.
His team’s results from the massive computational simulations have revealed exactly how the calcium binds to the fusion peptides. The researchers have found that the calcium helps the fusion peptide attach itself to the outer surface of the membrane, enabling a specific part of the peptide to wiggle its way into the thick of the membrane and anchor it there.
“The calcium ions are positively charged, and the tops of the lipids that compose the membrane are negatively charged,” says Weinstein. “The thick of the membrane has no charge, it’s hydrophobic’ — it ‘dislikes’ water. That’s why it welcomes the specific residues of the fusion peptides that are also hydrophobic and are found in the fusion peptides of the other coronaviruses that enter in the same way. And that’s why they can wiggle in.”
Without calcium this would not happen, he adds, and if you remove or replace these specific hydrophobic residues from the peptide, there won’t be any penetration even in the presence of calcium.
Getting ready to design inhibitors
The research kicked off after a group of virologists working in coronavirus research at Cornell University in Ithaca contacted Weinstein’s team in April. They needed help on a project aimed at better understanding the function of SARS-CoV-2. At that point, there wasn’t any specific molecular level structural information about this very dynamic and flexible region of the viral spike protein, or what happens to it during its interaction with the membrane. “That’s why this region had never been targeted for inhibition,” says Weinstein.
But with the latest results, he adds, it’ll now be easier for us and others to identify and test potential inhibitors — small molecules, including peptides. When the spike is ready for piercing the cell’s membrane with its fusion peptides, these inhibitors would cover the fusion peptide’s spearhead, preventing the spike from penetrating the membrane.
The scientists are now developing exploratory large-scale molecular dynamics simulation to identify the best binding sites for inhibitors. As they go along, they add new information about possible sites to the machine learning algorithms used to identify inhibitors. “We have just started collaborating with experts in the application of artificial intelligence to drug discovery at IBM — the team headed by Payel Das,” says Weinstein.
The supercomputer they’ve used, AiMOS, has greatly helped to speed up the compute-intensive simulations process. “AiMOS is the 29th fastest supercomputer in the world thanks to it’s hybrid GPU-CPU compute architecture,” says Chris Carothers, a Professor in the Computer Science Department at Rensselaer Polytechnic Institute and Director of the Center for Computational Innovations that manages the machine. Over the past seven months, he adds, Rensselaer and IBM have diverted over 150,000 compute-node hours to COVID-19 research projects — corresponding to “more than 100 years of compute time using a typical workstation.”
The results obtained from Weinstein’s study are also valuable on a wider scale. They are giving COVID-19 researchers worldwide a new set of structural and dynamic information about a region of the virus that is essential to understanding its function both outside and inside the cell.
Gradually, all the pieces of this jigsaw puzzle are coming together. Soon, we’ll know enough to keep the virus at bay — for good.
This article first appeared on the COVID-19 HPC Consortium blog