VR Used to Study Coronavirus Proteins

By Deborah Bright

Right now, thousands of scientists all over the world are working around the clock to address the current coronavirus outbreak. On Thursday, February 20, 2020, two of those scientists — one in Australia and one in America — met in a virtual room. A vast gray dome, the room’s walls were haphazardly littered with files: the latest scientific paper on the 2019 novel coronavirus’ spike (S) protein, a scanning electron microscopy (SEM) image of four severe acute respiratory syndrome (SARS) coronavirus particles, and protein models waiting to be loaded into a simulation.

The virtual room was inside Nanome software, a collaborative virtual reality (VR) software platform for molecular design. The two scientists — Dr. Michael Kuiper, a biomolecular modeler at Data61, the data science arm of Australia’s national science agency CSIRO, and Dr. Michael Bishop, a drug discovery specialist at Nanome — were there to talk about Dr. Kuiper’s protein modeling work on the novel coronavirus using VR. The conversation was recorded and can be viewed on YouTube.

“This isn’t the first coronavirus outbreak we’ve seen,” said Dr. Kuiper to Dr. Bishop. “We learned a lot from SARS.”

Dr. Kuiper’s virtual hands reached out and took hold of the SEM image showing the four viral particles. “This image of SARS shows a surface protrusion — that’s the spike protein. This protein is crucial for coronaviruses to attach to their host cells.”

Scientists like Dr. Kuiper are interested in how the novel coronavirus (2019-nCoV) uses its spike protein to attach to human cells. Because 2019-nCoV is evolutionarily related to SARS, researchers are investigating SARS’ point of entry — the receptor binding domain of angiotensin converting enzyme 2 (ACE2) — as the binding site for 2019-nCoV’s spike protein.

Like SARS, 2019-nCoV is a respiratory virus. It spreads through the air when an infected individual coughs or sneezes. Researchers have posited that when a susceptible individual inhales the viral particles, the particles attach to host cells via their spike protein in the individuals’ airways. When a virus enters a cell, it takes over the cell’s biological processes, turning it into a viral particle–producing factory and disrupting its normal function. The cells eventually die, slough off, and build up in the airways, making it difficult for the individual to breathe.

In order to understand this mechanism at the cellular level and think of possible ways to interrupt it, Dr. Kuiper has turned to virtual reality. “We want to understand the function of these proteins in a realistic environment,” said Dr. Kuiper. “In real life, proteins wiggle and jiggle. By loading protein models into Nanome software and running simulations, we can interrogate the binding sites, right down to the amino acid residues and hydrogen bonds, and come up with a model that can potentially be used to identify a therapeutic target.”

Scientists recently solved the structure of 2019-nCoV spike protein using cryo-electron microscopy (cryo-EM), but there were gaps. That’s something Nanome could help with. Dr. Kuiper loaded the cryo-EM 2019-nCoV spike protein structure and highlighted the missing residues. He then dragged his own model over and placed it on top.

“Molecular modeling can fill in the gaps,” he said. “This really highlights the power of VR, collaboration, and sharing data.”

Dr. Kuiper’s virtual finger touched a virtual button; ACE2 appeared, and the two domains shook and rotated in mid-air until they came together as one.

“Amazing,” Dr. Bishop said.

Dr. Kuiper and Dr. Bishop took turns passing the proteins back and forth, cycling through views — a surface view showing hydrophobicity, stick mode showing the hydrogen bonds, etc. Then, almost as though they were entering another dimension, they stepped into the structure.

“See this?” Dr. Kuiper asked, his virtual finger pointing to a now larger than life amino acid residue deep within the structure. “This model shows the same lysine and tyrosine residues as the SARS spike protein, but there are new amino acids and new hydrogen bond interactions, too. The authors of the recent paper measured the strength of this interaction at 10- to 20-fold higher than SARS. They speculate that this is why it may be so virulent this time around. Looking at the interface as we do in VR really helps us understand why that interaction is so strong. For example, there seems to be a number of new hydrogen bond interactions that were not present in SARS/ACE2: Y498Q–K493Q–S501N, and a new salt bridge between D406E and V417K.”

Viruses are constantly evolving. The more we know about these interactions, the more we can do with that knowledge. Not only can we use the knowledge to try to predict the next outbreak, we can also use it to develop novel therapeutics. For example, Dr. Kuiper is also running simulations using small molecules.

“We can load small molecule drugs — there’s one currently used on HIV — to see how they might inhibit coronavirus protease activity,” said Dr. Kuiper. “We put the protease in an aqueous environment, load in the small molecule, and allow the protein to go through its dynamic flexing. This can open up a lot of cryptic sites that aren’t otherwise available. It’s an iterative approach, though; if we find something that binds, and we have experimental evidence that indicates that it binds, then we can take it to the next level. We can use VR to see the residues, even pull them out as a separate molecule, interrogate the interaction, and modify them.”

Dr. Kuiper says he uses VR every day. “I was a skeptic, until about a year ago. It’s a great addition to what we do. It’s a tangible way for us to manipulate the molecules in a way that’s intuitive. And the level of detail is astounding — you see things that you miss in 2D.”

But you don’t have to take Dr. Kuiper’s word for it — you can see them too. Nanome software is free to download, and Dr. Kuiper encourages people to contribute to and improve upon his models. “For example, the spike protein is glycosylated but that hasn’t been included yet,” he said.

Near the end of the discussion, Dr. Kuiper told a story that illustrates the power and potential VR has to contribute to scientific progress:

“I was visiting my dad on a remote island in Australia, and I was showing some of his friends what I’ve been working on. I put on my headset and walked into a room belonging to someone 10,000 miles away, in New York. We ended up having a great conversation about protein modeling. Something like this could only happen in VR.”

The discussion ended with Dr. Kuiper pointing out that the visualizations in Nanome are snapshots of the simulation, like a few representative frames to tell the story of a movie. “The simulations are large — 600,000+ atoms — and include the spike protein, the target protein, and surrounding water. It can take weeks to simulate a few hundred nanoseconds of molecular movement.” But computers are getting faster. He sees a future in which his simulations are longer, and infinitely more data can be captured and used to halt the next outbreak in its tracks.