AI discovers oral medications can stop multiple coronaviruses

Scientists have developed a new oral antiviral drug using artificial intelligence and Hollywood animation software that can block infections from a variety of coronaviruses, including the SARS-COV-2 variant, SARS and MERS.

The compound, known as WYS-694, reduced viral load in infected mice by more than four times when orally administered, providing hope for preventing future pandemic outbreaks. Unlike existing treatments that target surface proteins that are prone to mutant, the drug is zero in hidden areas inside the virus that remain stable between coronavirus species.

The discovery comes from an abnormal collaboration between computational biologists, infectious disease experts and drug developers at the Wyss Institute at Harvard University, who borrow tools from the film industry to simulate the virus’s hijacking of human cells.

Movie magic encounters viral mechanics

Instead of studying static snapshots of viral proteins, the research team used Houdini (the same procedural animation software behind the movie’s special effects) to create a dynamic model of the transformation of coronavirus spike proteins during infection. This approach reveals crucial things that were missed by previous methods.

“We believe that when the virus initially binds to its host cells, the constant region remains hidden, but is accessible in a critical time window to prepare itself for membrane fusion,” explains Charles Reilly, the first author of the study and former WYSS lead scientist. “Targeting these could be a way to essentially lock the virus before it can be released into the host cell cytoplasm during the fusion phase.”

The team’s molecular dynamics simulations show that when the coronavirus spike protein binds to human cell receptors, mechanical tension causes a physical separation of a specific peptide sequence (TNFTISVTT) from the virus’s S2 subunit. This isolation exposes a hidden cavity containing highly conserved residues-areas without mutations, as they are crucial for viral function.

Key research results:

• WYS-694 is 12.5 times more potent than its predecessor compound
• Fight against SARS-COV-1, SARS-COV-2, MER and multiple Covid variants
• Virus entry unrelated to AXL kinase inhibition
• Demonstrate oral bioavailability and extend drug exposure
• Reduced viral load in infected mice by more than 4 times

From screen to laboratory bench

Using this mechanical insight, the researchers calculated that about 10,000 existing drugs were screened to find molecules that could be tightly bound to the newly identified pocket. The highest-ranked oral drug is Bemcentinib, a FDA-approved cancer drug that previously showed anti-rotation activity in clinical trials.

However, bemcentinib inhibits the action of AXL kinase, making it unclear whether its antiviral effect comes from this known mechanism or directly binds to viral proteins. To solve this problem, medicinal chemist Joel Moore designed structurally similar compounds that retain viral binding but lose AXL activity.

The resulting molecule WYS-633 demonstrated that antiviral activity is independent of kinase inhibition. “To make clear evidence that Bemcentinib achieves its antiviral activity through the mechanism we proposed and further enhances its efficacy, we need to create structurally similar compounds (analogs) that lack affinity for AXL for AXL but maintains its affinity with spike protein binding pockets,” Reilly said.

Mechanical force drives virus into

What is particularly novel about this study is that its focus is on mechanical forces associated with viral infection. The team’s simulations show that when the coronavirus spike protein binds to human cell receptors, the resulting tension triggers the conformational changes required for membrane fusion.

This mechanical deployment process seems crucial in multiple virus families. The researchers noted that similar force-dependent mechanisms play a role in influenza, HIV and other enveloped viruses, suggesting that their approach may have a wider application.

Through molecular dynamics simulations of advanced transduction, they demonstrated that WYS-694 stabilizes key structural elements that would otherwise be rearranged to form the post-fusion state of the virus. The drug is essentially a molecular “wrench” that prevents spike proteins from completing their shape-transforming sequences.

Beyond Covid: Broad Spectrum Protection

Tests show that WYS-694 blocked infection through multiple coronavirus variants, including Alpha, Beta, Delta, Gamma and Omicron strains. The compound also blocks the entry of SARS-COV-1 and MERS-COV, demonstrating true broad-spectrum activity.

This versatility stems from targeting conserved internal regions rather than variable surface domains recognized by current antibody therapies. Since these hidden regions do not face immune stress, their mutation frequency is much lower than the exposed binding sites.

The pharmacokinetic characteristics of WYS-694 also show promise for practical use. When taken orally in mice, the drug can prolong exposure, with a peak concentration of 24 hours, potentially allowing for convenient administration once a day.

Passed Advanced AI Verification

To further validate their findings, the researchers used Google’s Alphafold 3 machine learning algorithm to predict the protein structure of its drug compounds. When the S1 subunit is shifted, these predictions always position WYS-694 within the target area, supporting its mechanical assumption.

Interestingly, when Alphafold 3 was modeled with the complete peak protein of the S1 region, it did not place the drug at the target site, confirming that the mechanical displacement of S1 was exposed to the binding pocket. The computational verification is complete with their experimental observations.

From defense funding to pandemic preparations

The project began in spring 2020, when emergency support from the Defense Advanced Research Projects Agency (DARPA) was unfolding with the COVID-19 pandemic. The urgency of the crisis has led the team to initially focus on repurposing existing FDA-approved drugs rather than designing entirely new molecules.

This fast-responsive approach led to the identification of bemcentinib within months, demonstrating how computational and experimental pipelines accelerate drug discovery in healthy emergencies.

Why does this mechanical approach succeed in the face of other strategies struggle? Traditional antiviral development usually targets static protein structures, but viruses are inherently dynamic machines. By focusing on the mechanical transformation required for infection, the researchers identified intervention points hidden in conventional screening methods.

The pipeline for future outbreaks

The approach developed for the project extends the coronavirus. The same comprehensive approach – combining molecular dynamics, AI-based docking, evolutionary analysis and medicinal chemistry, may be targeted at other viruses that rely on membrane fusion entry.

“This approach has great potential for drugs targeting many other virus fusion proteins, including other virus families such as flu, HIV, Ebola, measles, etc.”

The researchers designed their computing pipelines to be modular and adaptable, able to incorporate new AI technologies when they emerge. This flexibility is crucial in dealing with future virus threats that may require rapid responses.

The road ahead

Although WYS-694 shows impressive preclinical results, important issues remain. The researchers acknowledged that they have not yet demonstrated direct binding to the predicted target sites through high-resolution structural studies. Such confirmation will require techniques such as cryoelectron microscopy or hydrogen exchange mass spectrometry.

The team also pointed out the limitations of the current computational approach, which prioritizes rapid hypothesis generation over detailed quantitative analysis. Future research could employ more sophisticated simulation methods to better understand the precise energy of drug binding and viral conversion.

However, the current results provide compelling evidence for novel antiviral compounds targeting mechanical aspects of the virus’s entry. As the research team pointed out, this represents “the first of new antiviral compounds that target membrane fusion and mechanical transformation of cells into the required viral S protein”.

For a world still fighting Covid-19 and working for future pandemic threats, the development of oral drugs with broad-spectrum coronavirus activity provides an invaluable complement to our antiviral arsenal. The ability to prevent infections, especially in areas where vaccination volumes are still limited, may be crucial to pandemic preparation.

AI, integration of physics-based modeling and traditional drug development demonstrates how interdisciplinary approaches address complex biological challenges. By borrowing tools from Hollywood and combining them with cutting-edge computing biology, these researchers have opened new avenues for antiviral discoveries that can reshape how we prepare for future infectious disease outbreaks.

Related