AI-Assisted Research Reveals Key Factor in Enhancing Antibody Effectiveness

Empowering antibodies to better activate the immune system

Antibodies are best known for their ability to latch onto and neutralize bacteria, viruses and other pathogens. But these immune proteins can do more than that: They also activate other components of the immune system, which then go to work to clear an infection. A new study from Scripps Research explores the factors that influence how effectively antibodies engage specific immune cells. Their results, described April 22, 2025 in Cell Reports, indicate that a higher ratio of antibodies to viral protein -- in this case, a part of HIV -- better engages two specific types of immune cells. While this discovery is most immediately applicable to experimental HIV vaccines, it has implications for the development of other drugs as well. "Many therapeutics and vaccines tap into antibodies' immune-stimulating function," says senior author Lars Hangartner, associate professor of immunology in the Department of Immunology and Microbiology at Scripps Research. "Understanding the mechanisms that govern it will help us to develop new versions that are better at their jobs because they have the most assistance from this secondary function." Hangartner's team employed an artificial intelligence (AI) tool for a key aspect of this research: designing altered versions of the HIV protein. This AI-based approach accelerated their work and could do the same for future projects. Fine-tuned by the immune system to latch onto a particular site on a pathogen, Y-shaped antibodies bind to their targets with their arms. Meanwhile, the Fc region (or "stem") of the antibody attaches to other immune cells. These include phagocytes, which engulf infected cells, and natural killer cells, which perforate and poison them. Previous research in Hangartner's lab and elsewhere has indicated that in many cases -- but not all -- this interaction bolsters antibodies' effectiveness and improves protection against infection. But scientists know little about what determines how well, if at all, the Fc region stimulates immune cells. Hangartner believed multiple factors could be at play, including binding location. Antibodies that target the same viral protein can bind to different spots on its surface, called epitopes. Likewise, the strength of the bond between the antibody and epitope can vary, as can the number of antibodies that bind near one another. To answer these questions in the new study, Hangartner's team focused on a well-studied target, HIV's Env protein, which the virus uses to invade human cells. But Env's epitopes are highly complex and so not suitable for the experiments they had in mind, which included relocating the epitopes on the viral protein to study the importance of binding location. Instead, they used a smaller, simpler, more accommodating epitope derived from the influenza virus. Even with the simpler flu epitope, conventional techniques for grafting it onto varying spots on Env would require considerable trial and error. What's more, the results would be limited by the protein's willingness to cooperate. An AI tool called AlphaFold2 helped them work around these challenges. With AlphaFold2, they designed Env proteins with the flu epitope placed exactly where they wanted it. They then selected and refined those designs, before expressing the altered proteins. They analyzed how relocating the epitope altered the behavior of two immune cells: natural killer cells and phagocytes. They also looked at how these two cell types responded when the antibodies formed weaker versus stronger bonds with the epitope. And finally, they evaluated how the cells reacted when one, two or three antibodies bound to a set of three Env proteins. Of these three factors, only one appeared to matter: The ratio of antibodies to Env. Both immune cells became most destructive when three antibodies bound per set of Env proteins. While the phagocytes displayed a low level of activity with a single antibody, natural killer cells had almost no response unless at least two were present. This research suggests that HIV vaccines could also enhance the response from these other immune cells, in addition to eliciting the immune system to produce antibodies against the virus. Experimental vaccines that produce antibodies that can bind at high ratios appear most likely to take advantage of this secondary effect. Antibody ratio appears to influence these interactions in other infectious diseases as well, according to Hangartner. "I think this rule is probably true for many other pathogens," Hangartner says, cautioning that only experiments will show whether such changes translate into better protection from disease. "It's likely but not a given." It's not clear yet if binding ratio's influence extends beyond infections to therapeutic antibodies, like those that target cancer cells or errant immune cells that promote inflammation. But, in general, Hangartner says, a better understanding of how these interactions activate other parts of the immune system could speed the development of these treatments.

How increasing antibody-to-antigen ratio enhances immune cell activation

Antibodies are best known for their ability to latch onto and neutralize bacteria, viruses and other pathogens. But these immune proteins can do more than that: They also activate other components of the immune system, which then go to work to clear an infection. A new study from Scripps Research explores the factors that influence how effectively antibodies engage specific immune cells. Their results, described in Cell Reports, indicate that a higher ratio of antibodies to viral protein -- in this case, a part of HIV -- better engages two specific types of immune cells. While this discovery is most immediately applicable to experimental HIV vaccines, it has implications for the development of other drugs as well. "Many therapeutics and vaccines tap into antibodies' immune-stimulating function," says senior author Lars Hangartner, associate professor of immunology in the Department of Immunology and Microbiology at Scripps Research. "Understanding the mechanisms that govern it will help us to develop new versions that are better at their jobs because they have the most assistance from this secondary function." Hangartner's team employed an artificial intelligence (AI) tool for a key aspect of this research: designing altered versions of the HIV protein. This AI-based approach accelerated their work and could do the same for future projects. Fine-tuned by the immune system to latch onto a particular site on a pathogen, Y-shaped antibodies bind to their targets with their arms. Meanwhile, the Fc region (or "stem") of the antibody attaches to other immune cells. These include phagocytes, which engulf infected cells, and natural killer cells, which perforate and poison them. Previous research in Hangartner's lab and elsewhere has indicated that in many cases -- but not all -- this interaction bolsters antibodies' effectiveness and improves protection against infection. But scientists know little about what determines how well, if at all, the Fc region stimulates immune cells. Hangartner believed multiple factors could be at play, including binding location. Antibodies that target the same viral protein can bind to different spots on its surface, called epitopes. Likewise, the strength of the bond between the antibody and epitope can vary, as can the number of antibodies that bind near one another. To answer these questions in the new study, Hangartner's team focused on a well-studied target, HIV's Env protein, which the virus uses to invade human cells. But Env's epitopes are highly complex and so not suitable for the experiments they had in mind, which included relocating the epitopes on the viral protein to study the importance of binding location. Instead, they used a smaller, simpler, more accommodating epitope derived from the influenza virus. Even with the simpler flu epitope, conventional techniques for grafting it onto varying spots on Env would require considerable trial and error. Furthermore, the results would be limited by the protein's willingness to cooperate. An AI tool called AlphaFold2 helped them work around these challenges. With AlphaFold2, they designed Env proteins with the flu epitope placed exactly where they wanted it. They then selected and refined those designs, before expressing the altered proteins. They analyzed how relocating the epitope altered the behavior of two immune cells: natural killer cells and phagocytes. They also looked at how these two cell types responded when the antibodies formed weaker versus stronger bonds with the epitope. And finally, they evaluated how the cells reacted when one, two or three antibodies bound to a set of three Env proteins. Of these three factors, only one appeared to matter: the ratio of antibodies to Env. Both immune cells became most destructive when three antibodies bound per set of Env proteins. While the phagocytes displayed a low level of activity with a single antibody, natural killer cells had almost no response unless at least two were present. This research suggests that HIV vaccines could also enhance the response of these other immune cells, in addition to eliciting the immune system to produce antibodies against the virus. Experimental vaccines that produce antibodies that can bind at high ratios appear most likely to take advantage of this secondary effect. Antibody ratio appears to influence these interactions in other infectious diseases as well, according to Hangartner. "I think this rule is probably true for many other pathogens," Hangartner says, cautioning that only experiments will show whether such changes translate into better protection from disease. "It's likely but not a given." It's not clear yet whether the binding ratio's influence extends beyond infections to therapeutic antibodies, like those that target cancer cells or errant immune cells that promote inflammation. But in general, Hangartner says, a better understanding of how these interactions activate other parts of the immune system could speed the development of these treatments.