During their post-docs, Elizabeth Wood, PhD, CEO of Jura Bio and Julie Norville, PhD, the company’s CSO, shared a vision to build a drug development company based on the principle of translating the success they were seeing in the development of therapies for malignancies to the autoimmune space. Over the past six years, the two scientists have combined their deep knowledge of artificial intelligence with synthetic biology, to uncover the area of T-cell receptor and antigen binding, and build an end-to-end discovery and development platform to develop immune-based therapeutics.
“Therapeutic success hangs on the TCR-Epitope-HLA synapse,” noted Cameron Gardner, PhD, Jura Bio’s director of research, during a talk at the Bioprocess International meeting, taking place this week in Boston.
To achieve their goal, the company needed a map from any antigen/HLA to any candidate TCR. The map would allow them to find TCRs that are, effective, safe, tunable, and polyclonal. In order to do that, the map would need to predict 1028 interactions (a conservative estimate that is still more than the number of stars in the universe.) A key bottleneck in mapping the synapse was the ability to manufacture and test TCRs, epitopes, and HLAs.
The company’s ML-first workflow allows it to build and train models from huge amounts of data, enabling Jura Bio to propose, build, and physically assay candidates at a very large scale.
Some groups work with small libraries with a large proportion of useful targets. Other libraries are large, but low quality. Jura Bio’s approach of bringing their technology to the manufacturing floor resulted in a large library, with a large number of useful targets.
“In areas where conventional methodologies are insufficient, machine learning emerges as the catalyst that unlocks the full potential of cell-based therapy and personalized medicine,” said Norville.”
Jura Bio’s ML-improved gene synthesis technology has generated a library of 100 billion potential human and improved TCR candidates that allow for antigen-specific TCR discovery and development. The company has already discovered TCRs for prostate cancer and other neoantigen targets that have never been before recorded in key HLA-types.
One example of target discovery, Gardner described, occurred when the company was sent sequences of six patients with melanoma refractory to MART-1 specific adoptive cell therapy. They were asked, can you generate additional TCRs for those patients? Using their HLA-matched variational synthesis library of TCR candidates, Jura Bio found 300 candidate sequences and was able to identify 10 candidates for expansion and engineering. The company has also manufactured peptidome- and virodome-scale libraries of antigens to help shed light on the underlying mechanisms of autoimmunity that still evade understanding.
“AI-ML and multiplex libraries are useful tools individually, but when combined together they can generate remarkable synergy, known as ML-ML variational synthesis,” noted George Church, PhD, founder and scientific advisory board chair of Jura Bio. “This has the potential to generate billions to trillions-fold of potential candidates. The mastery of this strategy for TCR, MHC, and T-cell engineering at Jura Bio is without competition and in huge demand.”
Now, Jura Bio announced a research collaboration with Syena, a cell therapy product company and subsidiary of the genome writing company Replay, to develop TCR-based therapies.
“We’re thrilled to announce our first partnership with Replay and its product company Syena. This partnership validates our approach of using synthesized human T-cell repertoires to generate safe and effective libraries to discover antigen-specific TCRs at scale,” said Wood.“The human immune system is a powerful source of safe and effective immune receptors, and while one patient might lack a TCR necessary to fight cancer, it may be present in another. By leveraging machine learning to rewrite the gene synthesis process from the ground up, we can produce extraordinarily high-quality immune receptors libraries to discover and train probabilistic machine learning models to ensure a faster development process that identifies TCRs recognizing the most challenging therapeutic targets.”
“Jura Bio’s highly differentiated TCR discovery platform has transformative potential and combines the power of synthetic biology with that of machine learning. We look forward to working with them to identify high-performing TCRs recognizing challenging but therapeutically important targets such as KRAS G12D,” said Adrian Woolfson, executive chairman, president, and co-founder of Replay. “Jura Bio’s unique tools and expertise offer a differentiated approach for discovering challenging TCRs of therapeutic significance. This will be invaluable in helping Replay and its cell therapy product company Syena to advance novel TCR-NK therapies into the clinic.”
Financial terms of the agreement were not disclosed. Jura Bio will receive an upfront payment as well as research funding for the period of the partnership. If the option is exercised, Replay and Syena will be responsible for global development and hold exclusive worldwide commercialization rights on all TCR-NK therapies resulting from the partnership.
In addition to the partnership with Replay and Syena, Jura Bio also announced $16.1M in financing. The funding will accelerate the mapping of the human adaptive immune system. Jura Bio’s goal is to have completed a predictive map of TCR-antigen-HLA binding powered by an off-the-shelf library of >100B synthesized human T cells and their cognate antigens and HLAs by the end of 2024 as well as to have expanded their machine-learning backed gene synthesis into the design and discovery of B cell receptors.