Nature has provided not only a synthetic machinery that can be used to accelerate the development of medicines (our long-term goal) but also a plethora of examples for how this machinery synthesizes medicines. However, the potential of polyketide assembly lines remains virtually untapped by medicinal chemistry. Beyond manipulating the DNA encoding these synthases and identifying suitable heterologous hosts, an incorrect understanding of the logic of these molecule factories has thwarted their engineering. Over the last several years, bioinformatic evidence has mounted that the modular unit recombined during assembly line evolution differs from the traditional polyketide synthase module that most scientists employ in their designs. Our lab has helped redefine the module such that a gatekeeping ketosynthase (KS) domain is at its most downstream position and has demonstrated that synthases designed with the updated boundary outperform those designed with the traditional boundary. After many design-build-test cycles, we are now able to rapidly engineer pentaketide synthases that produce preparative levels of stereochemically-dense polyketides from E. coli. Our lab is positioned to further our knowledge of assembly line logic as we engineer assembly lines that generate medicinally-relevant products. Through Specific Aim 1 (the bottom-up approach) we will push the substrate tolerance limits of KSs, asking them to accept intermediates with substituents beyond the b-carbon that differ from those they naturally accept. Through 3 ligations with DNA encoding 5 pikromycin modules, 125 pentaketide synthases will be constructed. Mass spectrometry methods, including imaging, will quickly identify struggling synthases. Guided by a bioinformatics/structural study of KS gatekeeping recently completed in our lab, we will predict what mutations will remove bottlenecks in these assembly lines. Gain-in-function mutants will inform future engineering. Through Specific Aim 2 (the top-down approach) pikromycin modules will be combined through 4 ligations to yield 100 heptaketide synthases. The products will be similar to narbonolide, the product of the pikromycin synthase, but with differing combinations of ketide units at the second, third, fifth, and sixth positions. After optimizing synthases as in the first aim, desosamine biosynthesis/transfer genes will be supplied to generate narbomycin analogs. As from the seminal, modular syntheses of macrolides performed by the Andrew Myers lab, we anticipate discovering several new macrolide antibiotics. In Specific Aim 3 (the horizontal approach) a library of 32 hybrid pentaketide synthases will be constructed using modules from the pikromycin and spinosyn assembly lines. We hypothesize that many of these will be inactive due to incompatibilities between KS and acyl carrier protein (ACP) domains at intermodular junctions. An interface repeatedly identified by docking servers for cognate KS and ACP domains will guide KS surface mutations to restore function to inactive synthases. We seek to identify a set of mutations that permit the docking of diverse ACPs, thus facilitating the recombination of all modules and providing access to as much polyketide chemical space as possible.

Modular polyketide synthases are enzymatic assembly lines that naturally synthesize many important human medicines and drug leads. If their synthetic abilities were harnessed, medicines could be developed much more rapidly. Through a new understanding of the polyketide synthase module and new technology developed by our lab, we are able to rapidly engineer highly active synthases and use them in the development of much needed antibiotics.