AI Meets Immunology:
Reimagining Personalized Medicine
By Anna Holt
Photography by Kat Lawrence
Imagine that health care professionals could predict how a patient’s immune system would respond to a treatment before the medicine ever reached the clinic. Envision vaccines with fewer side effects or cancer therapies precisely tailored to each patient’s unique biology.
A collaboration between UNC Charlotte professor of chemistry Kirill Afonin and researchers from the National Cancer Institute and National Center for Advancing Translational Sciences, Marina Dobrovolskaia and Alexey Zakharov, respectively, is bringing this reality closer than ever. Together, they have developed AI-Cell (Artificial Intelligence-Cell) — a first-of-its-kind tool that mimics how human immune cells respond to RNA- and DNA-based nanomedicines. Their innovation has the potential to revolutionize gene therapy, making it safer and more personalized.
“Essentially, AI-Cell is a computational algorithm that thinks like the human immune cell, which will make it possible, ultimately, to better match treatments to patients with certain diseases based on their own biology,” Afonin explained.
This breakthrough addresses a significant challenge in gene therapy: while nucleic acids in medicines and vaccines can boost the body’s natural defenses, they also carry the risk of triggering harmful immune responses. Take, for example, the rapid development of COVID-19 vaccines using mRNA technology. While incredibly effective, these vaccines were tested extensively to ensure they didn’t provoke severe reactions. Similarly, cancer patients undergoing chemotherapy often face debilitating side effects because their immune systems overreact to treatments. AI-Cell offers a way to predict and potentially prevent such issues.
Afonin’s team has joined forces with experts in biology, machine learning and immunology to further diversify and improve the capability of this AI-Cell platform. Prominent among them is UNC Charlotte’s Brittany Johnson, as assistant professor of biological sciences, whose research is designed to gain understanding of pathogen responses to host environments and identify novel therapeutic points of intervention during infection of organs and tissues.
AI-Cell is a digital library of known nucleic acid nanoparticle combinations — and associated immunological responses — freely available to biomedical researchers. With this data, scientists can better predict which compositions, structures, shapes and amounts of nanoparticles will work best to curb disease without causing extremely negative immune responses.
“Essentially, AI-Cell is a computational algorithm that thinks like the human immune cell, which will make it possible, ultimately, to better match treatments to patients with certain diseases based on their own biology.”
Kirill Afonin
Picture nucleic acid nanoparticles or NANPS, a term coined by Afonin, as Lego blocks. Each NANP is a known structure, yet the options to arrange these blocks into various forms are virtually unlimited.
“If designed correctly, your body will recognize these artificially made NANPs as its own components, which can initiate and guide various biochemical processes and help fix the problem from within,” explained Afonin.
Johnson added, “Often, we think the immune response is always productive. It is a little bit more like a see-saw, and while you want things to be in balance, you need the immune system to peak at a certain time. And then, you want that to resolve, so that the body can go through the healing process.”
Afonin-Johnson lab research focuses on the ways nucleic acid nanoparticles — or NANPS — can be assembled into various shapes and injected into cells to communicate with the innate immune system — the body’s first line of defense against pathogens.
Through this “molecular language,” the level of immune response is determined — and is designed to improve and personalize the safety and efficacy of vaccines, immunotherapies and drug delivery.
Because nucleic acids in medicines or vaccines effectively “piggyback” on the immune system’s fight against an invading virus or bacteria, there is a risk of instigating a severe response that ultimately is not healthy.
"The big idea is to develop a molecular language to explain to our bodies and immune system how to reveal and deal with certain diseases — and make this technology user-friendly, widely available, personalized and affordable," Afonin said.
Delivering targeted genetic material to cells — with programmable messages and nanoparticles capable of controlling or reducing adverse side effects — is a particularly vexing proposition.
“Our bodies look for specific sequences and aspects of foreign nucleic acids. They are looking for something amiss,” Johnson said. “Because of that, when we deliver an intervention like nucleic acids, you have to be cautious because you could stimulate an unwanted immune response.”
Brittany Johnson
In addition, there are near-limitless structures and combinations possible with nanoparticle technology.
“Particularly for nucleic acid-based nanotechnology, AI-Cell is going to be incredibly helpful because you can move in both directions. Because a patient is diagnosed with a certain disease, you may be trying to get a certain immune response. You can use this tool. You can say, ‘Okay, I know what immune response I want — now I need to go backward: What nucleic acids can I use to get there,’” explained Johnson. “Instead of starting with unlimited possibilities, we’re engineering it to this targeted portion.”
To grasp the impact of AI-Cell and the advancement in treating diseases on the cellular level, it helps to understand the basics of biotechnology and nanotechnology targets.
DNA is the building block or blueprint of the human cell, and RNA decodes the blueprint.
In the context of gene therapies, NANPs work in concert with the body's natural response to pathogens and disease. NANPs are effective because nucleic acids in drug delivery can augment the body's natural defense and self-healing processes.
Knowing if the body needs to control or lower inflammation or stimulate more action from the immune cells is a balancing act. Combining rationally designed and lab-constructed NANPs with the AI-Cell tool aids drug and vaccine design towards this goal.
There is huge potential to offer new ways to treat a broad spectrum of malignancies, from cancers to cardiovascular problems and infectious diseases. This includes therapeutics, diagnostics and preventative biomedicine, such as vaccines with mRNA technology.
Using human blood immune cell data in combination with AI-Cell’s predictive power renders a more realistic view of likely patient immune responses
In 2022, the National Cancer Institute featured the Afonin Lab’s work with immunostimulatory NANPs on the cover page of a Congressional presentation.
Afonin, who has logged the immunorecognition of close to 180 NANPs, said, "What we're doing will be of huge help to the scientific community and the growing field of RNA nanotechnology — the field right now is starting to bloom. AI-Cell has already become a very dynamic platform, which has no limits for expansion.
AI-Cell is the result of more than 10 years of research using expertise in the growing fields of biotechnology and artificial intelligence.
Undergraduate and graduate students in the Afonin Lab, located in the Klein College of Science, are working on two fronts to advance the AI-Cell platform. They are investigating and experimenting to find new NANPs to examine how they work, and they are contributing to the dataset behind AI-Cell so that other labs can "test" or query the data.
Funding for AI-Cell and related research comes from the National Institute of General Medical Sciences of the National Institutes of Health with additional support from NCI and NCATS.
A forthcoming manuscript from Afonin and Johnson will explore AI-Cell’s capability to predict the immunostimulation of NANPs in the human central nervous system. This work was done in collaboration with NCATS, Ball State University and UNC Charlotte.
UNC Charlotte alumna Morgan Chandler ’21 Ph.D. is a former member of the Afonin Lab. Following graduation, she completed a 13-month fellowship at the U.S. Food and Drug Administration. Now, she works as a scientist at Mimetas, a small biotech company based just outside Washington, D.C.
Around the time Chandler started her current role, a peer-reviewed paper about AI-Cell (for which she served as lead author) was published in Small, a leading nanoscience and nanotechnology journal. In it, the research team offers an in-depth explanation of the approach — the first of its kind — to use “state-of-the-art transformer neural networks to predict immunological activity and thus advance the current understanding of the NANP properties.”
Chandler noted that during her time as a Ph.D. student, the Afonin Lab “compiled this huge database to identify all the structural parameters and biological activities of our NANPs. That was the library we passed on to the National Center for Advancing Translational Sciences lab to build the predictive model using all their computational expertise.”
Her work at UNC Charlotte added to the AI-Cell database in testing the 2-D and 3-D structures that are now being used by current Afonin Lab student researchers. And soon, that number of students who are involved in expanding AI-Cell could increase.
Afonin plans to submit a $3 million funding proposal to the National Science Foundation to start a new center called Traineeship in the Advancement of RNA Nanotechnologies. Through this center, Charlotte faculty would train University graduate students to work with NANPs and carry out cutting-edge research in this field, with a particular emphasis on NANP immunotherapies.
“The predictive capacity of AI-Cell is just going to increase,” Afonin said. “It’s very, very promising, but there’s more work to do for sure. It’s also very exciting because our team will continue working on it. There are still a lot of questions to answer.”
Anna Holt is a freelance writer based in Charlotte, North Carolina.