Journal of Molecular Biology
A Comparison of Directed Evolution Approaches Using the β-Glucuronidase Model System
Introduction
The properties of proteins can be altered through site-directed mutagenesis1., 2., 3., 4., 5., 6., 7., 8., 9., 10., 11. or directed evolution (also called in vitro, experimental or laboratory evolution).12., 13., 14., 15., 16., 17., 18., 19., 20., 21. The latter approach does not require an understanding of protein structure, but is predicated upon a battery of evolutionary techniques. Molecular diversity is generated either by random mutagenesis of a protein-coding gene22., 23., 24. or chimeragenesis of two or more genes.25., 26., 27. Libraries of mutant genes are expressed in populations of microorganisms. Clones exhibiting improvements in a desired property are isolated in high-throughput screens or selections. Selected clones are often further mutated and/or randomly recombined for the next round of expression and selection.
Directed protein evolution remains as much an art as a science. Practitioners often disagree about the most effective way to generate molecular diversity. High-throughput screens and selections are generally re-invented for each application. Experiments are seldom repeated, even though changes in parameters such as mutation rate, population size and selection stringency can potentially alter the outcome. Here, we present side-by-side comparisons of different methods for generating molecular diversity and high-throughput screening.
We previously directed the evolution of Escherichia coli β-glucuronidase (GUS) variants with β-galactosidase (BGAL) activity.28 We utilized a strong inducible expression system and a simple high-throughput screen based on visual comparison of colonies stained with 5-bromo-4-chloro-3-indolyl-β,d-galactopyranoside (X-gal). After three rounds of DNA shuffling (random recombination),29 we isolated a GUS variant (called clone 1.3.1 to indicate experiment 1, round 3, clone 1) that exhibited a 500-fold improvement in catalytic efficiency (kcat/KM) in reactions with para-nitrophenyl-β,d-galactopyranoside (pNP-gal). Additional rounds of DNA shuffling or random mutagenesis and screening did not lead to variants with increased BGAL activity, despite numerous attempts.28
Adaptation in this experiment might have stopped after three rounds for several reasons. First, the best enzymes might have reached the upper limit of the dynamic range of the X-gal-based screen. Second, the non-quantitative and imprecise nature of the X-gal screen might have gratuitously reduced the genetic diversity within the population. Third, the benefits of DNA shuffling might have been offset by costs that further decrease the sequence diversity within a library: (A) DNA shuffling often generates a small subpopulation of very fit recombinants,18., 21., 28. and thus causes tight population bottlenecks. (B) The basic procedure that we employed is associated with a high rate of random mutation,30 so that many advantageous sequence combinations might have been masked by deleterious mutations. (C) The mutation bias of DNA shuffling likely reflects the strong transition bias of Taq polymerase,23 so that transversion mutations were not sampled.
Here, we develop a versatile and relatively inexpensive high-throughput assay system, and employ the GUS system to compare mutagenesis and screening strategies. The outcome of each evolution experiment was contingent upon the combination of methods utilized. The new screen led to the identification of a wider variety of beneficial mutations. To our surprise, however, these beneficial mutations drove each other into extinction when DNA shuffling was not employed. These results explain why DNA shuffling and other recombination methods enhance the efficiency of directed evolution.
Section snippets
Development of a semi-automated screen
The objective of this study is to employ different methods to direct the evolution of GUS into a BGAL, and to compare the outcomes. The efficiency of any high-throughput selection or screen is a function of its throughput, sensitivity, precision and dynamic range. We considered several selection strategies. In vivo selections are very high in throughput (>108 clones/round), but are generally insensitive. The BGAL activity of the wild-type GUS28 is many orders of magnitude less than that
Discussion
We have directed the evolution of gusA variants with BGAL activity using: (1) X-gal-agar plates and DNA shuffling (sexual recombination; 1.3.1), (2) and (3) pNP-gal microplates and recursive mutagenic PCR (asexual reproduction; 2.10.1, 3.5.1), and (4) X-gal agar plates and combinatorial cassette mutagenesis (high-frequency sexual recombination; 4.1.1–4.1.11). Post-game analysis of the gusA sequences (genotype) and pNP-gal activities (fitness) shows that different screening and
Materials
All materials, including the His6-tagged (but otherwise wild-type) gusA gene, were obtained as described.54 The Mutazyme was from Stratagene (La Jolla, CA); DNA sequencing kits were from Perkin-Elmer/Applied Biosystems (Foster City, CA). The Multidrop384 microplate dispenser, Multiskan Ascent spectrophotometer and Assist microplate stackers were from Thermo Labsystems (Waltham, MA). The silicone microplate seals were from Specialty Silicone Products†. The environmental rotator
Acknowledgements
L.R. performed most of the experiments. O.A. developed the high-throughput assay, and repeated the recursive random mutagenesis experiment. M.G. performed the combinatorial cassette mutagenesis and screening experiment. We thank the National Science Foundation (MCB0109668) for support, Richard Lenski for his ideas on clonal interference, and the other members of the Matsumura group for discussion. We thank the late Alec Hodel for reading the manuscript; this work is dedicated to his memory.
References (58)
- et al.
Engineering enzyme specificity
Curr. Opin. Chem. Biol.
(1998) - et al.
Replacement of Asp333 with Asn by site-directed mutagenesis changes the substrate specificity of Escherichia coli adenylosuccinate synthetase from guanosine 5′-triphosphate to xanthosine 5′-triphosphate
J. Biol. Chem.
(1994) - et al.
Guided evolution of enzymes with new substrate specificities
J. Mol. Biol.
(1996) - et al.
The effect of high-frequency random mutagenesis on in vitro protein evolution: a study on TEM-1 beta-lactamase
J. Mol. Biol.
(1999) - et al.
In vitro evolution of beta-glucuronidase into a beta-galactosidase proceeds through non-specific intermediates
J. Mol. Biol.
(2001) - et al.
High-throughput screening of enzyme libraries
Curr. Opin. Biotechnol.
(2000) - et al.
Cost-effective whole-cell assay for laboratory evolution of hydroxylases in Escherichia coli
J. Biomol. Screen
(2001) - et al.
Cassette mutagenesis: an efficient method for generation of multiple mutations at defined sites
Gene
(1985) Histochemical, spectrophotometric, and fluorometric GUS substrates
- et al.
Directed evolution of beta-galactosidase from Escherichia coli by mutator strains defective in the 3′→5′ exonuclease activity of DNA polymerase III
FEBS Letters
(2001)