Identifying Reactivity Bridgeheads in Sequence Space by Ultrahigh-Throughput Screening in Microdroplets

Existing biodiversity is an untapped reservoir of functional molecules that by far outweighs the number man-made systems known. Yet we know little about its functional potential, because information about the metagenome of uncultivated organism is difficult to access and interpret, chiefly for two reasons: (i) massive sequencing data exists, but cannot be extrapolated to functional predictions, when truly novel enzymes (without precedent) are targeted; and (ii) functional assays of metagenomic material typically yield very few hits, so that screening campaigns for rare activities are expensive or impossible to carry out. Use of picoliter droplets resolves this bottleneck by scale-down of the volume of functional assays (using picoliter droplets as reaction compartments) and scale-up of the number of experiments (to >107 precision assays in a day: > three orders of magnitude more than achievable in robotic facilities). We have recently demonstrated such a screen on an unprecedented scale and successfully identified a large number of new enzymes from environmental DNA that could not have been predicted by bioinformatics and have to be considered rare. This suggests that a wide range of reactions - slow and fast, rare and frequent - can be discovered. This new approach to functional metagenomics and biocatalyst discovery provides fundamentally new types of insight into the understanding of biocatalysts: Currently most enzyme annotation is based on sequence similarity - this work (and many previous other studies) suggest that this annotation is misleading. The functional tests in picodroplets provide data that complement and correct sequence analysis. We show that activities can be targeted that have never been explored by metagenomics before. Most previous work has focused on esterases/lipases, for which many sequences and structural and functional data are available. This suggests that we can access 'new' sequence space. The locations of our present hits in sequence space are far apart, i.e. genuinely new (and not just homologs of well-known enzymes). This suggests that they (and those from future experiments) will provide entirely new bridgeheads for functional annotation. Consequently our understanding of the functional proteome will be propelled beyond sequence similarity annotation. This approach is uniquely powerful for 'unpredictable' reactions (for which so few experimentally-characterised examples exist that sequence comparison is not even an option). These are shown to be 'promiscuous' secondary reactions of enzymes. Enzyme promiscuity has recently received much attention, because promiscuous enzymes are now seen as evolutionary starting points that, e.g. after gene duplication, can be further refined by adaptive evolution and lead to neofunctionalised proteins. The more starting points we know, the more new activities we can evolve - and the better can we understand where to find functions in the vastness of sequence space. Here we will use 'mechanistic baits' that call for a particular mechanistic arrangement: for example a hydrophobic pocket, an active nucleophile, a cofactor etc. to explore sequence space by reactivity, i.e. with the interest of an organic chemist. Elicit specific types of reactivity from environmental DNA as source material will enable new enzyme chemistry of unnatural reactions and tell us where in sequence space 'virginal' environmental microbiota find promiscuous 'head start' activities.