Dengue virus, comprising four serotypes (DENV1-4), initiates an immune response upon infection, leading to the production of antibodies, the body’s defense mechanism. These serotypes are distinguished by their immunodominant antigen protein E, which forms the virus’s outer membrane, providing specificity for antibody binding. Each serotype elicits a unique antibody response, forming an immunological fingerprint and contributing to the antibody repertoire. Understanding this repertoire aids in vaccine development by targeting antigens that induce protective antibodies. In the event of a secondary infection, pre-existing antibodies afford faster protection. Dengue infection can lead to severe dengue fever, potentially fatal in certain cases. Despite numerous vaccine development efforts, including those aimed at alleviating symptoms in naive individuals, there remains a lack of specific therapeutics for dengue virus. Broadly neutralizing antibodies, generated in response to a wide range of antigens, show promise as a potential therapy.
In this experiment, Natalie and colleagues aimed to gain a comprehensive understanding of the immune response. They immunized mice with DENV serotypes containing complex protein E variants, including the whole protein E of DENV2 (Whole E-2) and the domain III of protein E from each serotype (EDIII-1 to EDIII-4). To establish long-term immunity against DENV, researchers focused on studying B cells, specifically long-lasting plasma cells derived from mouse bone marrow. Fluorescence-activated cell sorting was employed to differentiate these cells effectively. Subsequently, the genome of the isolated plasma cells was analyzed using 10X Genomics technology, known for its high-quality cell characterization. next-generation sequencing was then utilized to unveil the genetic sequences of antibody-producing proteins, thereby sequencing the antibody repertoire. Furthermore, recognizing the potential to efficiently curate datasets based on prior samples, the team leveraged machine learning techniques. Combining machine learning with network theory, a computational approach incorporating mathematical and graphical methods, facilitated a deeper understanding of the complexity inherent in the antibody repertoire. To ensure the accuracy and precision of their findings, the researchers created three distinct machine-learning training sets: Random Forest (RF), multilayer perceptron, and support vector machines. Among these, RF emerged as particularly adept at addressing their research requirements.
Following the implementation of RF, the findings from the investigation into immune response, focusing on protein E alterations, suggest that the utilization of antigens containing the entire protein E leads to a more diverse antibody repertoire compared to immunization with EDIII antigens. The latter, being less complex, exhibits a narrower range of antibodies in comparison. Additionally, it has been uncovered that the heavy chain, as opposed to the light chain, plays a more significant role in antigen binding, influenced by the structure of antibodies, particularly the length of the complementarity-determining region 3 (CDR3) and the amino acid composition, notably the presence of type Y amino acids. Consequently, this research concludes that dengue infection correlates with the CDR3 length and amino acid composition of antibodies.
In summary, scientists are endeavoring to uncover the antibody repertoire to better comprehend the antigen-specific response of antibodies, with the ultimate goal of developing effective therapeutic interventions. Despite the complexity inherent in the antibody repertoire posing a challenge to its elucidation, integrating computational analyses into machine learning has enabled scientists to swiftly acquire data concerning the properties of myriad antibody types and interpret them. Based on this evidence, it can be argued that technological innovations hold promise for the treatment of virus-based diseases.
Author: Tuğçe Çayır
Editor: Elif Duymaz
Reference: E N Natali, A Horst, P Meier, V Grieff, M Nuvolone, L M Babrak, K Fink , and E Miho (2024). The dengue-specific immune response and antibody identification with machine learning. Npj vaccines, 9(16). https://doi.org/10.1038/s41541-023-00788-7
– Bioinfocodes Scientific News Service –
News articles prepared by our team members, reviewing and compiling scientific research published in journals with an impact factor greater than 20 (click here for the list).