A systematic bioinformatics approach to characterise antibody properties

Abstract

Antibodies are crucial components of the humoral immune responses, and their therapeutic potential is gaining attention. For antibodies to be effective, it is essential that they exhibit specificity towards antigens, mediate biological functions and are administered at the correct dosage. Given the numerous parameters influencing specificity, function, and optimal concentration, assessing what constitutes an effective antibody involves a multidimensional approach. In this thesis, we explored the suitability of the systems serology methods for our datasets and adapted a systems antibody approach to analyse monoclonal antibody characterisation data. Our findings revealed that while the systems serology was not the ideal analysis option for the serum data available, the systems antibody approach provided insights into the relationships between the features of the antibodies.

Popular Abstract

Antibody with a heart of gold

When our bodies encounter a pathogen for the first time, our immune systems learn how to eliminate it, and if we emerge victorious from that fight, our prize is not only a healthy recovery but also ammunition for the next fight. This ammunition is called antibodies. Antibodies are proteins capable of recognising a pathogen and telling the other components of our immune systems how to get rid of it. A recovered patient would have these antibodies circulating in their serum. Transferring that serum to a sick patient affected by the same pathogen is called “serum therapy”. The ones who have watched the true-story-based Disney movie “Togo” will know that serum therapy has been around for a while. The movie tells how Leonhard Seppala and his sled dog Togo heroically deliver serum to the diphtheria-stricken Alaska town of Nome in 1925. In real life, the heroism of serum therapy took a toll after the 1930s when a new hot therapy called antibiotic therapy seized the scene in history. However, this fame came with a price: antibiotic resistance. It turned out that pathogens could also learn and transfer knowledge, thus becoming resistant to antibiotic therapies. Remembering the success of serum therapies, we again turned our face to them. However, this time around, instead of administering serum directly, without exactly knowing the components and possibly introducing side effects, technology allows us to design targeted antibody therapies where we can select which antibodies we want to transfer. But which antibodies do we want to transfer?

The question of what makes an antibody good might sound easy at first. It simply needs to be able to fight off the pathogen. However, we cannot try every antibody on every patient and see which works. Therefore, we need to find antibody features that indicate a successful antibody. This is not straightforward either. Remember how Togo was initially described as undersized, troublesome, and untrainable in the movie, but he turned out to have a heart that made him successful? Same with antibodies, we might be looking at how well they bind to a pathogen, which part of the pathogen they bind to, if they can immediately stop the pathogen from entering cells; and even though all these features are important, another hidden key feature might be the answer to successfully eliminating the pathogen.

The project we have been working on applies a multidimensional, data-driven approach to systematically analysing antibodies to find the “successful” ones. Similar approaches are prominent in analysing serum samples, called systems serology. We have tried adapting the data analysis methods used in systems serology to analyse antibodies, which we called systems antibody analysis. We first designed a data analysis application with user interface where researchers can input experimental and inherent data on different antibodies and get some plots showing the relationships between antibody features and group antibodies with similar features. Then, using the application, we analysed data from antibody experiments.

The correlation analysis investigated the relationship between the antibody features and whether these features correlate with each other. For example, we wanted to see if antibody A was better at binding the pathogen than antibody B, did that also mean that A would promote more phagocytosis (an immune response where antibodies call other immune cells to “eat up” the pathogen)? Or if A had a longer structure than B, would that mean that it could recruit more help from other immune components? Our results showed that even though there were some trends, it wasn’t always easy to reach a solid conclusion. The clustering part of the application tried to group similar antibodies together. We input the analysis all the features of the antibodies we had data for, and the antibodies were clustered based on these features. This revealed that different groups of antibodies with different functional profiles could be clustered. For example, a cluster was better at calling other immune system components; the next cluster was better at promoting phagocytosis. Overall, the project showed us the intricate nature of antibody analysis, the depth of understanding we can achieve, and the potential it holds for future therapies. Is there one exceptional antibody that will cure the pathogen? Or do we need to combine different antibodies from different groups? Which antibodies have a heart of gold like Togo? These are the questions that can be answered with further research.

Master’s Degree Project in Bioinformatics 60 credits 2024
Department of Biology, Lund University

Advisor: Pontus Nordenfelt (Associate Professor)
Department of Clinical Sciences, Infection Medicine

• author: Yanartas, Cagla Defne
• supervisor: Pontus Nordenfelt
• organization: Degree Projects in Bioinformatics
• course: BINP52 20232
• year: 2024
• type: H2 - Master's Degree (Two Years)
• subject: Biology and Life Sciences
• language: English
• id: 9175571
• date added to LUP: 2024-09-27 12:40:23
• date last changed: 2024-09-27 12:40:23