Periodic Reporting for period 4 - PicoCB (Exploring the Chemical Biology of Sequence Space via Picoliter Droplets)
Reporting period: 2021-04-01 to 2022-09-30
Directed evolution of functional proteins has arguably emerged as an approach to protein engineering that can complement or better design-led approaches to protein function. However, as a random process, enormous numbers of variants have to be screened and selected to have a chance to identify successful catalysts. This process is costly and cumbersome: Industrial screening facilities require investment of tens to hundred millions of dollars. In this project quantitative biological experiments are scaled down to an extremely miniaturised format, where only picoliter volumes are needed for one ion and ultrahigh throughput screening consequently becomes much cheaper. Millions of variants can be screened in monodisperse oil-in-water compartments ('microdroplets’) that are generated at kHz frequencies in microfluidic devices. The droplet compartment constitutes a link between a given phenotype and its encoding genotype, by capturing reaction product, and thus providing a unique system to screen for catalysis. 'Fitness landscapes' composed of characterisation of more than a million individual library members give insight into the quality of a library, allowing a quantitative assessment of proteins that fulfil a desired function more of less well. These quantitative fitness landscapes provide the basis for selection of the best library members for a function: e.g. the interconversion of members of enzyme superfamilies along the lines of catalytic promiscuity, understanding the factors governing specificity and the mechanistic interpretation of the observed evolutionary pathways. We have applied this screening system of unprecedented capacity for directed evolution and metagenomic screening of enzymes in in vivo and in vitro formats. This system allows large scale experiments that would not be possible with conventional, lower throughput approaches.
We have advanced the project in several areas:
(i) Screening of metagenomic libraries for rare and promiscuous activities that characterise environmental gene collections for their reactivity and potential for applied biocatalysis.
We have expanded the range of assays for which droplet screening is possible: this list now includes glycosidases, carboxyesterases, redox enzymes and proteases. Several practical improvements have made the screening process easier: improvements of signal/noise ratios by clonal amplification in droplets, development of in vitro workflows, improvements to droplet characteristics (stability and leaking properties) to implement assays on longer timescales and new device designs.
(ii) Developing a fundamental understanding of enzyme evolution based on fitness landscapes that record data on multiple, promiscuous activities
We have tested the effect of single mutations on multiple activities in two groups of the promiscuous alkaline phosphatase superfamily to probe this hypothesis, quantifying the effect of site-saturating mutagenesis of an analogous residues. Statistical analysis suggests that no physico-chemical characteristics alone explain the mutational effects, but are dominated by their structural context. Likewise the effect of changing the catalytic nucleophile itself is not reaction-type specific. Mapping of “fitness landscapes” of four activities onto the possible variation of a chosen sequence position revealed tremendous potential for re-specialization by single point mutations, highlighting catalytic promiscuity as a powerful predictor of adaptive potential: a multifunctional catalyst may act as a springboard for efficient functional adaptation.
(iii) New libraries with insertions and deletions
Insertions and deletions (InDels) are among the most frequent changes observed in natural protein evolution, yet their potential has hardly been harnessed in directed evolution experiments. We have set up a method (TRIAD), a simple and efficient Mu transposon mutagenesis approach for generating libraries of single InDel variants with one, two or three triplet nucleotide insertions or deletions. The libraries give access to ~90% transposon insertion positions, so that >10e5 variants per insertion library can be achieved. Fitness analysis of InDel libraries of phosphotriesterase obtained via TRIAD revealed that InDels are 5- to 10-fold more deleterious than substitutions. However, directed evolution by screening these libraries for improved promiscuous arylesterase activity yielded over 80 distinct and substantially improved variants. Kinetic characterization revealed weaker trade-off usually observed in evolution with substitution variants. Overall, our results suggest that InDel libraries created using TRIAD provide ready access to functionally distinct beneficial variants and expand the directed evolution toolbox toward functional trajectories that would not arise from substitution mutagenesis libraries.
(iv) Evolution of gene networks to build up signalling networks in vitro.
We have developed and implemented droplet-based kinase assays that allow us to screen libraries for the transfer of a phosphate group, involving one or multiple enzymes. This assay was applied to ultrahigh-throughput screening of docking interactions in synthetic MAPK cascades. Since thus far no high-throughput method to monitor this reaction had been available, the advent of a generalizable platform to screen for phosphorylation within cascades, suggests that this class of reactions can now be explored by directed evolution.
(i) Screening of metagenomic libraries for rare and promiscuous activities that characterise environmental gene collections for their reactivity and potential for applied biocatalysis.
We have expanded the range of assays for which droplet screening is possible: this list now includes glycosidases, carboxyesterases, redox enzymes and proteases. Several practical improvements have made the screening process easier: improvements of signal/noise ratios by clonal amplification in droplets, development of in vitro workflows, improvements to droplet characteristics (stability and leaking properties) to implement assays on longer timescales and new device designs.
(ii) Developing a fundamental understanding of enzyme evolution based on fitness landscapes that record data on multiple, promiscuous activities
We have tested the effect of single mutations on multiple activities in two groups of the promiscuous alkaline phosphatase superfamily to probe this hypothesis, quantifying the effect of site-saturating mutagenesis of an analogous residues. Statistical analysis suggests that no physico-chemical characteristics alone explain the mutational effects, but are dominated by their structural context. Likewise the effect of changing the catalytic nucleophile itself is not reaction-type specific. Mapping of “fitness landscapes” of four activities onto the possible variation of a chosen sequence position revealed tremendous potential for re-specialization by single point mutations, highlighting catalytic promiscuity as a powerful predictor of adaptive potential: a multifunctional catalyst may act as a springboard for efficient functional adaptation.
(iii) New libraries with insertions and deletions
Insertions and deletions (InDels) are among the most frequent changes observed in natural protein evolution, yet their potential has hardly been harnessed in directed evolution experiments. We have set up a method (TRIAD), a simple and efficient Mu transposon mutagenesis approach for generating libraries of single InDel variants with one, two or three triplet nucleotide insertions or deletions. The libraries give access to ~90% transposon insertion positions, so that >10e5 variants per insertion library can be achieved. Fitness analysis of InDel libraries of phosphotriesterase obtained via TRIAD revealed that InDels are 5- to 10-fold more deleterious than substitutions. However, directed evolution by screening these libraries for improved promiscuous arylesterase activity yielded over 80 distinct and substantially improved variants. Kinetic characterization revealed weaker trade-off usually observed in evolution with substitution variants. Overall, our results suggest that InDel libraries created using TRIAD provide ready access to functionally distinct beneficial variants and expand the directed evolution toolbox toward functional trajectories that would not arise from substitution mutagenesis libraries.
(iv) Evolution of gene networks to build up signalling networks in vitro.
We have developed and implemented droplet-based kinase assays that allow us to screen libraries for the transfer of a phosphate group, involving one or multiple enzymes. This assay was applied to ultrahigh-throughput screening of docking interactions in synthetic MAPK cascades. Since thus far no high-throughput method to monitor this reaction had been available, the advent of a generalizable platform to screen for phosphorylation within cascades, suggests that this class of reactions can now be explored by directed evolution.
We have developed and used a microfluidic screening platform in which water-in-oil droplets, a man-made alternative to the genotype-phenotype linkage provided by the cell, are used as 'evolutionary units'. The monodisperse droplets are produced in microfluidic chips so that quantitative and sensitive measurements can be performed. Ultrahigh-throughput sorting enables selection of droplets according to their optical properties. Using this format, ~10e8 biochemical reactions can be performed and analysed per day, typically in pico- to femtolitre volumes, resulting in dramatically lower costs when compared to robotic screening. We have employed such droplets to identify new functional proteins (binders and catalysts) from large libraries of mutants. Such capacity to find 'a needle in a haystack' provides a unique opportunity for exploration of protein sequdnce space that would be difficult or impossible with the current methodologies and we will use this approach to break new ground in the systematic understanding of biomolecular interactions.