Cytiva

Sofie Stille

Over the past three decades, the success of biotherapeutics research has unlocked new avenues to treat and prevent a wide variety of diseases, including cancer, autoimmune, and metabolic disorders. These therapeutics, produced in living cells contain high levels of host cell-derived contaminants. As a result, their purification process becomes crucial for boosting the efficiency of industrial manufacturing and ensuring the broad patient population’s safety.

The first step in the purification process centers around the capture step, relying heavily on the "golden tool" known as affinity chromatography. This method allows for over 95% purity in a single step, thanks to the targeted interaction between a ligand and the protein of interest.

The development of such specific ligands is in itself challenging.

In the 1970s, ligands were derived from naturally occurring binders, with Protein A—sourced from Staphylococcus aureus—for antibody purification serving as an excellent example. As antibody-based therapies expanded, greater emphasis was placed on efficient purification platforms.

In the 1990s, a decade-long effort enabled the cloning and recombinant expression of Protein A for large-scale, controlled production. Moving into the 21st century, sequences of modified ligands resulted in improved performance. This approach was further refined through the use of phage display libraries, utilizing relevant proteins as scaffolds to create vast ligand libraries with random amino acid sequences. Screening of these libraries led to more selective, optimized ligands. This innovative approach is today's standard and was awarded the Nobel prize in 2018. This powerful technology does not require any target information but relies on the random ligand generation and the iterative screening process of large libraries which is very labor intensive and time consuming. Indeed, it is not unusual for the development of a novel chromatographic solution to take two to three years; half of that time is dedicated to the ligand discovery phase.

Today, a paradigm shift is needed in purification tool development to speed up the process dramatically. Cytiva tackles this challenge by applying an AI rational ligand design approach, transforming the ligand identification phase.

Leveraging the latest breakthroughs in molecular recognition and structure prediction. (e.g., RoseTTAFold, AlphaFold-2, ESMFold, OmegaFold), Cytiva’s innovative approach is driven by a Machine Learning Artificial Intelligence (MLAI) software. This AI-powered tool predicts a target's 3D structure with remarkable accuracy, based solely on its amino acid sequence. Combined with generative diffusion models (e.g. RFdifussion), the algorithm enables rapid virtual identification of in silico ligands. Selected de novo ligands are synthesized and tested in a laboratory setting using traditional techniques.

We’re confident this AI approach is a game changer in the development of novel affinity resins, substantially reducing labor and time. Once established, we envision, the time to market being under one year. This approach promises higher-quality ligands with better selectivity, stability, and productivity, unlocking the potential to address not only blockbuster treatments, but also a wider range of targets and orphan therapies. 

In summary, AI will revolutionize the design and production of affinity ligands, enabling novel drugs to reach a broader patient population more swiftly.