A bold breakthrough signals a potential reshaping of mpox defenses that could extend beyond the current outbreak. With the aid of artificial intelligence, an international collaboration has taken a meaningful early step toward sharper, more accessible vaccines and antibody therapies for monkeypox virus (MPXV). The virus can trigger severe pain and, in the worst cases, death, with the greatest risk to children, pregnant individuals, and people with weakened immune systems. In findings published in Science Translational Medicine, researchers show that mice mounted strong neutralizing antibody responses after receiving a viral surface protein that AI helped pinpoint. This outcome points to a promising direction for future mpox vaccines or antibody-based treatments.
The mpox outbreak of 2022 swept through multiple countries, infecting more than 150,000 people and causing nearly 500 deaths. Health authorities relied on smallpox-based vaccines to shield the most vulnerable, but these vaccines are costly and complex to manufacture because they employ an entire weakened virus.
"Unlike a whole-virus vaccine that is large and intricate to produce, our approach centers on a single protein that’s much easier to manufacture," explained Jason McLellan, a professor of molecular biosciences at The University of Texas at Austin and co-lead author of the study.
Discovering Powerful Antibodies From Patients
The study’s other lead authors, Rino Rappuoli and Emanuele Andreano at the Fondazione Biotecnopolo di Siena in Italy, identified 12 antibodies capable of neutralizing MPXV. They achieved this by analyzing blood from individuals who had recovered from the virus or who had previously been vaccinated. While these antibodies were clearly active, the team did not yet know which parts of the virus they targeted.
MPXV presents a variety of surface proteins, and at least one is essential for the virus to spread. Some newly discovered antibodies interfered with this process, but scientists hadn’t identified which surface protein was responsible. To design effective vaccines or therapies, they needed to map the exact pairing between antibody and viral protein, the so-called antigen.
AI Reveals a Key Viral Protein That Had Been Overlooked
To crack the antibody–antigen pairing, McLellan’s group employed the AlphaFold 3 model to predict which of roughly 35 MPXV surface proteins would bind strongly to the patient-derived antibodies. The model highlighted a protein named OPG153 with high confidence, and subsequent lab experiments confirmed the prediction. This result suggests that OPG153 could serve as a valuable target for antibody-based therapies or for a new vaccine design that prompts the immune system to attack mpox more effectively.
"Without AI, it would have taken years to identify this target," said McLellan, who also holds the Robert A. Welch Chair in Chemistry and helps lead Texas Biologics, a UT Austin initiative focused on therapeutic innovation. "It was thrilling because no one had ever considered OPG153 as a vaccine or antibody target. It had never been shown to be a target of neutralizing antibodies."
Because MPXV is closely related to the smallpox virus, these findings may also inform improved vaccines or treatments for smallpox, a disease of particular concern due to its transmission and mortality potential.
Toward Next-Generation Vaccines and Antibody Therapies
The team is now refining versions of the antigen and antibodies to improve effectiveness, reduce cost, and simplify production compared with current strategies that rely on weakened poxviruses. Their long-term aim is to evaluate mpox and smallpox vaccine antigens and antibody treatments in human trials. McLellan describes their approach as a form of "reverse vaccinology."
"We started with people who survived monkeypox infection, isolated the antibodies they produced naturally, and traced back to what part of the virus acted as the antigen for those antibodies. Then we engineered the antigen to elicit similar antibodies in mice," McLellan explained.
UT Austin has filed a patent application for the use of OPG153 (and its derivatives) as a vaccine antigen. The Fondazione Biotecnopolo di Siena has filed patents for antibodies that target OPG153.
Other UT Austin contributors include Emily Rundlet, Ling Zhou, and Connor Mullins. Funding for this work came, in part, from the Welch Foundation.
Contemplating the implications, this research invites discussion: should we pursue antigens that enable targeted vaccines over traditional, whole-virus approaches, even if such a shift requires new regulatory paths? And as AI-assisted discovery accelerates vaccine design, how should we balance speed with rigorous safety testing to ensure broad public trust? Share your thoughts and predictions in the comments.
