Published on Web 07/17/2002 In Vitro Selection for Catalytic Activity with Ribosome Display Patrick Amstutz,† Joelle N. Pelletier,†,‡ Armin Guggisberg,§ Lutz Jermutus,†,⊥ Sandro Cesaro-Tadic,† Christian Zahnd,† and Andreas Plu¨ckthun*,† Biochemisches Institut, UniVersita¨tZu¨rich, Winterthurerstrasse 190, CH-8057 Zu¨rich, Switzerland, and Organisch-Chemisches Institut der UniVersita¨tZu¨rich, Winterthurerstrasse 190, CH-8057 Zu¨rich, Switzerland Received February 8, 2002 Abstract: We report what is, to our knowledge, the first in vitro selection for catalytic activity based on catalytic turnover by using ribosome display, a method which does not involve living cells at any step. RTEM-â-lactamase was functionally displayed on ribosomes as a complex with its encoding mRNA. We designed and synthesized a mechanism-based inhibitor of â-lactamase, biotinylated ampicillin sulfone, appropriate for selection of catalytic activity of the ribosome-displayed â-lactamase. This derivative of ampicillin inactivated â-lactamase in a specific and irreversible manner. Under appropriate selection conditions, active RTEM-â-lactamase was enriched relative to an inactive point mutant over 100-fold per ribosome display selection cycle. Selection for binding, carried out with â-lactamase inhibitory protein (BLIP), gave results similar to selection with the suicide inhibitor, indicating that ribosome display is similarly efficient in catalytic activity and affinity selections. In the future, the capacity to select directly for enzymatic activity using an entirely in vitro process may allow for a significant increase in the explorable sequence space relative to existing strategies. Naturally occurring enzymes catalyze a wide variety of selection methods sample the entire library in a single experi- chemical reactions and are increasingly used in pharmaceutical, mental step. Such experiments require a direct coupling of the industrial, and environmental applications as a result of their phenotype, which is to be selected for, and its encoding genetic high reactivities and specificities. However, the direct improve- information, the genotype. ment of biocatalysts remains challenging, and the yet more For the selection of enzymatic activities, three general ambitious goal of developing enzymes with new catalytic approaches can be distinguished: methods performed entirely functions still seems almost elusive. Although our knowledge in vivo, those working “partially” in vitro, and those performed of structure-function relationships of enzymes has significantly completely in vitro.1 In the in vivo approach, a genetic library increased, rational protein design is still a difficult task, encoding enzyme variants is transformed into cells where the especially for improved catalysis. The recently developed variants are expressed and selection takes place. Because strategy of directed evolution can be used as a complement to selection protocols are generally based on a growth advantage, rational design. In directed evolution, a protein function of e.g. complementation of an auxotrophy or resistance to a interest is evolved in the laboratory by mimicking Darwinian cytotoxic compound,5-8 in vivo selection of catalysis is limited evolution in multiple successive rounds of diversification (library to those activities giving rise to a growth advantage. Moreover, generation) with subsequent selection or screening (reviewed microbial genomes have evolved to deal with environmental in refs 1-3). selection pressure. The expression host can therefore be surpris- Screening, which involves the analysis of single protein ingly “creative” in escaping selection pressure, such as by higher variants, can be automated for high-throughput protocols but expression of a poor catalyst or the use of alternative metabolic remains laborious and time-consuming, therefore limiting the pathways, completely by-passing the enzymatic activity that is size of libraries which can be handled.4 As opposed to screening, the actual target of selection. Partially in vitro methods can offer an alternative to in vivo * Corresponding author. Tel. (+41-1) 635 55 70. Fax: (+41-1) 635 57 12. E-mail: [email protected]. methods. In these methods, the library is also introduced into † Biochemisches Institut, Universita¨t Zu¨rich. cells, resulting in display of the protein of interest, usually on § Organisch-Chemisches Institut der Universita¨t Zu¨rich. the surface of phage,9 bacteria, or yeast.10 Selection then occurs ‡ Present address: De´partement de Chimie, Universite´ de Montre´al, C. P. 6128, Succursale Centre-ville, Montre´al, PQ, Canada. ⊥ (5) Orencia, M. C.; Yoon, J. S.; Ness, J. E.; Stemmer, W. P.; Stevens, R. C. Present address: Cambridge Antibody Technology, The Science Park, Nat. Struct. Biol. 2001, 8, 238-242. Melbourn, Cambridgeshire SG8 6JJ, U.K. (6) Altamirano, M. M.; Blackburn, J. M.; Aguayo, C.; Fersht, A. R. Nature (1) Griffiths, A. D.; Tawfik, D. S. Curr. Opin. Biotechnol. 2000, 11, 338- 2000, 403, 617-622. Retraction, Nature 2002, 417, 468. 353. (7) Crameri, A.; Raillard, S. A.; Bermudez, E.; Stemmer, W. P. Nature 1998, (2) Olsen, M.; Iverson, B.; Georgiou, G. Curr. Opin. Biotechnol. 2000, 11, 391, 288-291. 331-337. (8) MacBeath, G.; Kast, P.; Hilvert, D. Science 1998, 279, 1958-1961. (3) Soumillion, P.; Fastrez, F. Curr. Opin. Biotechnol. 2001, 12, 387-394. (9) Dunn, I. S. Curr. Opin. Biotechnol. 1996, 7, 547-553. (4) Joo, H.; Lin, Z.; Arnold, F. H. Nature 1999, 399, 670-673. (10) Mendelsohn, A. R.; Brent, R. Science 1999, 284, 1948-1950. 9396 9 J. AM. CHEM. SOC. 2002, 124, 9396-9403 10.1021/ja025870q CCC: $22.00 © 2002 American Chemical Society In Vitro Selection of a Displayed Enzyme ARTICLES in vitro, that is, outside the cell, allowing fine-tuning of the selection pressure and selection conditions. Both entirely in vivo or partially in vitro techniques require a transformation step, limiting the applicable library size to cellular transformation efficiencies. Typically, transformation efficiencies of 107-108 cells/µg DNA for yeast and 109-1010 cells/µg DNA for Escherichia coli (E. coli)11,12 are achievable. A considerable amount of work arises from the need of ligating and transforming such a library after each round of randomiza- tion. Furthermore, cytotoxic proteins cannot be displayed at all. Technologies working completely in vitro can overcome these limitations, because no living cells are involved at any step. Two approaches that are carried out completely in vitro can be distinguished. In one, a compartmentalization of individual variants is achieved in water-in-oil emulsions.13 Since the in Figure 1. Principle of ribosome display selection for catalytic activity. situ detection of fluorescent products and direct optical sorting DNA encoding the protein of interest is transcribed and translated in vitro. of such droplets, harboring the catalytic proteins, have not yet Because the mRNA carries no stop codon and translation is stopped with been reported, the physical link between genotype and pheno- high Mg2+ concentrations, stable ternary complexes of mRNA (black), type is still required, and the emulsions have to be broken up ribosome (blue) and tethered nascent protein (red) are formed. These can be used directly for selection with a biotinylated suicide inhibitor. Excess before an affinity selection or sorting can be performed. The inhibitor is removed by gel filtration, and the labeled complexes are captured other approach makes use of the concomitant presence of mRNA with avidin coated magnetic beads. After washing to remove any untrapped and nascent protein at the ribosome during an in vitro translation ribosomal complexes, the selected complexes are destroyed to elute the 14 mRNA. Finally, reverse transcription (RT) and PCR are used to amplify for coupling of genotype and phenotype. Here, the most the genetic information of the selected clones. prominent techniques are ribosome display;15,16 mRNA display, 17 - 18 also termed “in vitro virus” or “mRNA protein fusion”; and active enzyme was enriched relative to less active mutants and 19 “ribosome-inactivation display system”. Below, we will relative to penicillin binding protein. The elution of covalently describe the first application of ribosome display to the selection trapped clones from the matrix to which they had been linked of catalytic activity in an effort to increase the library sizes is generally achieved by cleaving within a linker region, either accessible beyond those in previously published accounts with chemically between the suicide inhibitor and the affinity tag or in vivo or partially in vitro methods. enzymatically between the displayed protein and the phage. To select catalysts in vitro, catalysis must be correlated or Even though entirely in vitro display techniques have great coupled to binding. Although reversible binding of enzymes to potential for selection of catalytic activity, this has not previously transition state analogues has been used frequently in the been reported. We believe that ribosome display is particularly 20,21 selection of novel catalysts, a more direct selection for well suited to applications based on mechanism-based inhibition, enzymatic turnover can be achieved by the use of mechanism- because ribosome display offers an elegant solution for elution 22 based (or “suicide”) inhibitors. These compounds are substrate of the selected genetic information. Because genotype and analogues that, upon turnover, are converted into a reactive phenotype are coupled