Researchers at the Agile BioFoundry (ABF) and UC Berkeley have demonstrated a high-throughput platform that leverages polyketide synthase enzymes to produce commodity chemicals and precursors to novel materials.
Polyketide synthases (PKSs) are enzymes found naturally in bacteria and fungi that can function as a flexible chemical factory for making a wide variety of molecules. Their modular structure allows them to be reprogrammed to produce chemicals not found in nature.
But engineering custom PKSs that can produce specific products can require years of trial and error, mostly because the “design rules” for these complex enzymes aren’t well understood and there isn’t a high-throughput platform for assembling and expressing them.
“We set out to change the current state of the art for PKS engineering,” said Jay Keasling, Professor of Chemical & Biomolecular Engineering at UC Berkeley and co-lead of the project. “Our lab had developed several tools to engineer PKSs and have used them to produce commodity and specialty chemicals and biofuels. The ABF had a complementary set of tools to increase the speed and reliability of engineering biological systems. This combination held the promise of creating a game-changing technology platform for engineering PKSs.”
The teams developed a high-throughput automated platform for PKS gene construction that involves PCR amplification, plasmid assembly, and sequence verification through next generation sequencing (NGS). The researchers made several optimizations that increased the platform’s success rate from 3% to 81%. The platform can rapidly build large amounts of PKSs — one researcher can build hundreds of PKSs in just three weeks.
To demonstrate its effectiveness, the platform was used to produce the commodity chemical valerolactam, as well as derivatives of this chemical that can serve as precursors to novel materials.
The researchers also developed machine learning tools for use in PKS design. Using the Automated Recommendation Tool, the researchers developed a machine learning algorithm that can predict successful PKS designs. The algorithm — trained on data from the researchers’ demonstration of valerolactam production — predicts the catalytic activity of PKS designs based on four distinct features that are computed by the high-throughput platform.
The algorithm accurately predicted production levels, which means researchers could use this to prioritize specific PKS designs, before spending time on building them.
The collaboration also made updates to ClusterCAD, an publicly available online platform that makes it easier to design and test PKSs for synthetic biology applications. These updates include an expanded database, powerful search tools, and other new features that will help users explore more chemicals that can be produced through PKS engineering.
“This work was greatly enabled by the knowledge and tools of the ABF,” Keasling said. “This work serves as a model for future engineering of biology to create novel chemicals and materials.”
ABF Capabilities used in this collaboration: