Protein engineering: Beating the odds.

Millions of novel protein sequences are currently being discovered, at an incredible pace. For instance, a recent article1 described the identifi cation of 8 complete and 789 draft genomes within a single metagenome library from an aquifer water sample. Unfortunately, the analysis of this huge pool of sequences remains challenging, given that the function of ~30–40% of the genes remains unknown and that initial annotations have oft en been proven to be incorrect. Similarly, protein engineering eff orts using directed evolution create large mutant libraries (commonly in the range of 106–109 variants) that must be interrogated to identify desired improved proteins and, in particulaar, enzymes. Consequently, the availability of rapid, reliable screening methods for the functional characterization of natural and man-made diversity remains the primary bottleneck in this fi eld. Two publications now report novel tools and concepts (Fig. 1) to address this challenge, going signifi cantly beyond the previous state of the art. In this issue, Chen et al.2 developed a technology platform they dubbed “μSCALE.” First, millions of protein variants produced by bacteria (Escherichia coli) or yeast (Saccharomyces cerevisiae) were spatially separated and mixed with magnetic microparticles in a microcapillary array. Th ese variants were monitored by a camera, allowing up to 10,000 samples to be distinguished per second. Th e hits were then retrieved by a very smart and precise laser-based extraction system (Fig. 1, bottom left ). Th is laser technique has the major advantage of isolating viable cells, allowing subsequent verifi cation of initial hits through the necessary genotype-phenotype linkage. Th e authors demonstrated the usefulness of this system for three important areas of application for proteins: the identifi cation of (i) a new antibody against a clinical target, (ii) a novel orange fl uorescent protein for biosensor applications and (iii) an improved alkaline phosphatase with reduced inhibition. Colin et al.3 describe a microfl uidics technology to interrogate a metagenome library (>106 variants) to identify novel enzymes with promiscuous activities, such that they can catalyze more than one distinct chemical reaction4. Th eir system is composed of monodisperse waterin-oil picoliter droplets, each, statistically, presumed to contain one E. coli cell PROTEIN ENGINEERING