Understanding how cells experience bioprocess conditions remains a major challenge in biomanufacturing. At QBDC Austria, Karlheinz Landauer and his team is addressing this problem by combining epigenetics with AI and machine learning. The multidisciplinary experts use DNA methylation as a “process historian” to assess cell fitness and uncover the impact of process perturbations such as leachables in single-use systems. The long-term goal is to enable manufacturing that is more robust, predictable, and transferable.
For Karlheinz Landauer the idea to use epigenetic signals as a window into how cells experience bioprocess conditions did not emerge from a long-term research plan. It came from a chance encounter. In 2018, after attending the BIO-Europe Spring event in Vienna, Landauer was waiting at a metro station when he struck up a conversation with another conference attendee, recognisable by his badge. “He told me about epigenetic characterisation of cells, by measuring the level of DNA methylation,” Landauer recalls. DNA methylation functions as a molecular switch that can turn genes on or off without changing the actual DNA sequence. In response to the environment, the level of methylation in the cell changes. “With this approach, he could measure cell fitness. It was the first time I had met someone using epigenetics to extract functional information from cells in vitro.”
QBDC
Landauer has many years of experience in bioprocessing. He earned a PhD in biotechnology in 2002 and worked many years for different pharmaceutical companies in various countries. In 2016 he decided it was time to set up a consultancy for the pharmaceutical industry, QBDC (Quality Biotech Development & Cells) in Switzerland. QBDC helps customers streamline their bioprocesses to reduce costs and improve quality. “At the BIO-Europe Spring event I was looking for partners, but I came home with a new idea,” says Landauer. “I realised that epigenetic signals add an extra layer of molecular-level insight into what’s happening inside the cell. It can add in our understanding how to drive cells toward desired outcomes and what are the causes of ‘mystery shifts’ in cell behaviour. Subtle variability in (single-use) bags and tubing, media and feeds can be part of those causes.”
EDDI algorithm
Landauer founded a separate company for the new technology, QBDC Austria, to enable external investment. The company uses DNA methylation to build AI and machine-learning models that capture how cells respond to environmental signals, such as contaminants. Changes in methylation patterns are read via microarrays and translated into actionable insights using the EDDI (Epigenetic Digital Data Interface) algorithm. “Comparing our epigenetic digital twin method to classical PAT and conventional digital twins you could say that we generate movies of the cell whereas the classical methods provide snapshots. Snapshots do not give by far that much information as movies.
At the molecular level, this approach enables faster root-cause identification. For example, we investigated a GMP seed train, which grew very slowly. Within six hours we discovered that the failure in growth was caused by a leachable. Using traditional wet-lab analysis it would have taken several months.”
Future
Landauer is proud of the status where they are now with their new method. “Things are starting to move now. We received an innovation award for EDDI and are already collaborating with corporations that seek our expertise in optimizing their clone selection process steps and value streams. This summer, we will introduce our first two scalable products and adopt our models across different species. A significant amount of exciting work lies ahead of us!”