Functional metagenomic discovery and characterisation of CAZymes by microfluidic methods
Repository URI
Repository DOI
Change log
Authors
Abstract
Enzymes are the engines of life and the ideal reagents for efficient, sustainable biocatalytic processes. However, the compendium of currently known enzymes does not cater for all desirable activities or exhibit the required properties. Therefore, the discovery and thorough characterisation of new biocatalysts is imperative to smoothen the path for a greener future. With the emergence of large-scale metagenomic sequencing projects, the number of protein sequences in databases has grown exponentially in recent years. This abundance of recorded sequence data, though, stands in contrast to the small number of experimentally verified functional annotation, which is crucial for the accurate assignment of enzyme activities. Functional metagenomics provides an alternative to the sequence-based exploration of the protein landscape. Experimental screening for new enzymes based on actual catalytic turnover is the most direct way to new biocatalytic function without relying on homology to already known sequences. However, traditional wet lab-based work for the functional discovery and characterisation of new enzymes is cumbersome and slow. The miniaturisation of assays in a microfluidic on-chip format, i.e. employing small water-in-oil droplets as reaction vessels that can be screened in ultrahigh throughput, has already allowed many standard lab procedures to be sped up, including functional screenings. Nevertheless, microfluidic techniques for functional metagenomic enzyme discovery studies are still in their early days. In this dissertation, I establish a protocol for the generation of plasmid libraries from soil samples as a resource for functional metagenomic studies and assess the size and quality of created libraires with nanopore sequencing. Moreover, I screen the SCV library, a million-membered metagenomic plasmid library, in microfluidic droplets via fluorescence-activated droplet sorting (FADS) for β-glucuronidase activity leading to the discovery of hit SN243. This enzyme belongs to the glycoside hydrolase (GH) family 3, a family that had no previous record of β-glucuronidase activity at the outset of this study. Detailed functional and structural characterisation provide evidence that SN243 is a genuine, efficient β-glucuronidase and promiscuous for other glycoside substrates. A wide-open active site cleft distinguishes the hit from otherwise homologous structures of GH3 members. The acquired data show that a functional metagenomic approach can shed light on assignments that are currently ‘unpredictable’ by bioinformatics. In general, genes coding for carbohydrate-active enzymes (CAZymes) are very often organised in clusters and the corresponding enzymes act together in the concerted degradation of large carbohydrate substrates. Therefore, it would be more desirable to screen metagenomic libraries with larger inserts (fosmids) and based on activity towards natural, rather than labelled model substrates with artificial leaving groups. To this end, I use an E. coli host that is genetically modified to express GFP from its genome and combine a classic growth experiment with microfluidic droplet screening via FADS. This novel approach uses the increase in fluorescence intensity caused by the multiplication of E. coli cells as a readout and allows screening with a mixture of natural oligosaccharides for CAZymes, while consuming only a few milligrams of the precious substrates in the entire screening campaign. With this assay format, five unique hits coding for up to eight CAZymes are discovered in one functional metagenomic campaign, and the activity of these predicted enzymes is verified with purified proteins. For the two hits harbouring the largest number of CAZymes, 1F12 (eight) and 1F7 (five), a comprehensive survey with natural polysaccharides isolated from plants is performed and their activity towards rhamnogalacturonan II and β-xylan/mixed linkage glucan, respectively, is demonstrated. While the acceleration of the functional discovery of biocatalysts with microfluidics is on the horizon, kinetic characterisation remains a bottleneck in most screening studies until today. To determine Michaelis-Menten kinetics, most researchers still pipet their different substrate dilutions by hand and measure the individual reactions in microtiter plates in a plate reader. This is not only tedious and slow, but also consumes a lot of plasticware, substrate and enzyme. I present a microfluidic platform that allows up to twelve full kinetic datasets to be determined with high accuracy, a large number of concentrations in droplets in parallel and within 30 minutes. Conveniently, this platform is based on an absorbance readout and, by changing the connected LED, can be easily adapted to different chromophores. In summary, the assays established in this thesis are demonstrated to be as reliable and sensitive as conventional plate-based assays but are faster, cheaper, and require much smaller amounts of chemicals. In the future, application of these protocols will lead to the more reliable and cost-efficient discovery of enzymes by function as well as their subsequent kinetic characterisation in high throughput.