Multiplex Next Generation Sequencing of Glycans and Nucleic Acids for Accelerating the Development of Cell Factories and Bioprocesses for Glycoprotein Biomanufacturing

Glycans consist of long branching chains of different sugars. They adorn proteins of all types. Some viruses use glycans to enhance the probability of infection. Some tumors hide from the immune system by using “stealth” glycoproteins on their surfaces. Most of the top drugs by revenue are glycoproteins, and their glycans affect drug safety and efficacy. This places a premium on making sure that their glycans are well-characterized and carefully managed. The complexity of glycan structures makes it difficult to control their structure on therapeutic drugs. Conventional glycan characterization methods have steadily advanced, but many challenges continue to hinder efforts to study and engineer these critical molecules. Here, an approach is being developed to rapidly and inexpensively sequence and quantify glycan structures. It will be first applied to accelerate the characterization and design of critical glycans required in biotherapeutics. A high-school outreach program on biological machine learning will introduce high school students to concepts underlying data science and its application to biological and biomedical questions. State-of-the-art technologies for glycan sequencing remain limited in their throughput and accessibility. They rely on methods with expensive, specialized equipment (e.g., mass spectrometry, NMR) or complex biochemistry (e.g., lectin arrays, exoglycosidase treatment). This research project aims to develop Glycosequencing, a technology that determines glycan structures using Next-Generation Sequencing (NGS) technologies. Using NGS to sequence and quantify DNA-barcoded lectins, Glycosequencing will measure a wide array of glycan features. The mapping of lectin binding patterns to glycan structures will be predicted using AI, trained on a large panel of recombinant glycoproteins with well-defined glycosylation patterns. This project will first identify the optimal set of lectins and biochemically characterize them. The lectin barcoding will be prepared, and lectin pooling will be optimized for NGS. To improve the AI accuracy, a training dataset will be built using 5 different recombinant proteins, transiently produced in a panel of >30 glycoengineered Chinese hamster ovary cell lines. To demonstrate the power of this technology, it will first be deployed to rapidly determine the structure of glycans on recombinant protein drugs. It will also be used to simultaneously profile glycans and the mRNA of the mammalian production host cells when cultured on 92 different media. This will allow the rapid characterization of the impact of culture conditions on protein glycosylation of monoclonal antibody drugs and a candidate hepatitis C vaccine. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.