Chapter 1
Introduction to Directed Evolution

1.1 General Definition and Purpose of Directed Evolution of Enzymes

Enzymes have been used as catalysts in organic chemistry for more than a century [1a], but the general use of biocatalysis in academia and, particularly, in industry has suffered from the following often encountered limitations [1b-d]:

• Limited substrate scope
• Insufficient activity
• Insufficient or wrong stereoselectivity
• Insufficient or wrong regioselectivity
• Insufficient robustness under operating conditions.

Sometimes, product inhibition also limits the use of enzymes. All of these problems can be addressed and generally solved by applying directed evolution (or laboratory evolution as it is sometimes called) [2]. It mimics Darwinian evolution as it occurs in Nature, but it does not constitute real natural evolution. The process consists of several steps, beginning with mutagenesis of the gene encoding the enzyme of interest. The library of mutated genes is then inserted into a bacterial or yeast host such as Escherichia coli or Pichia pastoris, respectively, which is plated out on agar plates. After a growth period, single colonies appear, each originating from a single cell, which now begin to express the respective protein variants. Multiple copies of transformants as well as wild-type (WT) appear, which unfortunately decrease the quality of libraries and increase the screening effort. Colony harvesting must be performed carefully, because cross-contamination leads to the formation of inseparable mixtures of mutants with concomitant misinterpretations. The colonies are picked by a robotic colony picker (or manually using toothpicks), and placed individually in the wells of 96- or 384-format microtiter plates that contain nutrient broth. Portions of each well-content are then placed in the respective wells of another microtiter plate where the screening for a given catalytic property ensues. In some (fortunate) cases, an improved variant (hit) is identified in such an initial library, which fulfills all the requirements for practical application as defined by the experimenter. If this does not happen, which generally proves to be the case, then the gene of the best variant is extracted and used as a template in the next cycle of mutagenesis/expression/screening (Scheme 1.1). This mimics "evolutionary pressure," which is the heart of directed evolution.

Scheme 1.1 The basic steps in directed evolution of enzymes. The rectangles represent 96 well microtiter plates that contain enzyme variants, the red dots symbolizing hits.

In most directed evolution studies further cycles are necessary for obtaining the optimal catalyst, each time relying on the Darwinian character of the overall process. A crucial feature necessary for successful directed evolution is the linkage between phenotype and genotype. If a library in a recursive mode fails to harbor an improved mutant/variant, the Darwinian process ends abruptly in a local minimum on the fitness landscape. Fortunately, researchers have developed ways to escape from such local minima ("dead ends") (see Section 4.3).

Directed evolution is thus an alternative to so-called "rational design" in which the researcher utilizes structural, mechanistic, and sequence information, possibly flanked by computational aids, in order to perform site-directed mutagenesis at a given position in a protein [3]. The molecular biological technique of site-specific mutagenesis with exchange of an amino acid at a specific position in a protein by one of the other 19 canonical amino acids was established by Michael Smith in the late 1970s [4a] which led to the Nobel Prize [4b]. The method is based on designed synthetic oligonucleotides and has been used extensively by Fersht [4c] as well as numerous other researchers in the study of enzyme mechanisms [4b]. This approach to protein engineering has also been fairly successful in thermostabilization experiments in which, for example, mutations leading to stabilizing disulfide bridges or intramolecular H-bridges are introduced "rationally" [5]. Nevertheless, in a vast number of other cases, directed evolution of protein robustness constitutes the superior strategy [6]. Moreover, when aiming for enhanced or reversed enantioselectivity, diastereoselectivity, and/or regioselectivity, rational design is much more difficult [3], in which case directed evolution is generally the preferred strategy [7]. In some cases, researchers engaging in rational design actually prepare a set of mutants, test such a "library" and even combine the designed mutations, a process that resembles "real" laboratory evolution, as shown by Bornscheuer and coworkers who generated 28 rationally designed variants of a lipase, one of them showing an improved catalytic profile [8]. Other examples are listed in Table 5.1 in Chapter 5. However, this technique has limitations, and standard directed evolution approaches are more general and most reliable.

Directed evolution of enzymes is not as straightforward as it may appear to be at this point. The challenge in putting the above principles into practice has to do with the vastness of protein sequence space. High structural diversity is easily designed in mutagenesis, but the experimenter is quickly confronted by the so-called "numbers problem" which in turn relates to the screening effort (bottleneck). When mutagenizing a given protein, the theoretical number of variants N is described by Eq. (1.1), which is based on the use of all 20 canonical amino acids as building blocks [2]:

where M denotes the total number of amino acid substitutions per enzyme molecule and X is the total number of residues (size of protein in terms of amino acids). For example, when considering an enzyme composed of 300 amino acids, 5700 different mutants are possible if one amino acid is exchanged randomly, 16 million if two substitutions occur simultaneously, and about 30 billion if three amino acids are substituted simultaneously [2].

Such calculations pinpoint a dilemma that accompanies directed evolution to this day, namely how to probe the astronomically large protein sequence space efficiently. One strategy is to limit diversity to a point at which screening can be handled within a reasonable time, but excessive diversity reduction should be avoided because then the frequency of hits in a library diminishes and may tend toward zero in extreme cases. Finding the optimal compromise constitutes the primary issue of this monograph. A very different strategy is to develop selection systems rather than experimental platforms that require screening. In a selection system, the host organism thrives and survives because it expresses a variant having the catalytic characteristics that the researcher wants to evolve. A third approach is based on the use of various types of display systems, which are sometimes called "selection systems," although they are more related to screening. These issues are delineated in Chapter 2, which serves as a guide for choosing the appropriate system. Since it is extremely difficult to develop genuine selection systems or display platforms for directed evolution of stereo- and regioselective enzymes, researchers had to devise medium- and high-throughput screening systems (Chapter 2).

1.2 Brief Account of the History of Directed Evolution

Scientists have strived for a long time to "reproduce" or mimic natural evolution in the laboratory. In 1965-1967 Spiegelman and coworkers performed a "Darwinian experiment with a self-duplicating nucleic acid molecule" (RNA) outside a living cell [9]. It was believed that this mimics an early precellular evolutionary event. Later investigations showed that Spiegelman's RNA molecules were not truly self-duplicating, but his contributions marked the beginning of a productive new area of research on RNA evolution as fueled by such researchers as Szostak, Joyce, and others [10]. At this point, it should be noted that directed evolution at RNA level is a very different field of research with totally different goals, focusing on selection of RNA aptamers, selection of catalytic RNA molecules, or evolution of RNA polymerase ribozyme and of ribozymes by continuous serial transfer [10]. The history of directed evolution in this particular area has been reviewed [10b, 11]. The term "directed evolution" in the area of protein engineering was used as early as 1972 by Francis and Hansche, describing an in vivo system involving an acid phosphatase in Saccharomyces cerevisiae [12]. In a population of 109 cells, spontaneous mutations in a defined environment were continuously monitored over 1000 generations for their influence on the efficiency and activity of the enzyme at pH6. A single mutational event (M1) induced a 30% increase in the efficiency of orthophosphate metabolism. The second mutational event (M2 in the region of the structural gene) led to an adaptive shift in the pH optimum and in the enhancement of phosphatase activity by 60%. Finally, the third event (M3) induced cell clumping with no effect on orthophosphate metabolism [12].

In the 1970s, further contributions likewise describing in vivo directed evolution processes appeared sporadically. The contribution of Hall using the classical microbiological technique of genetic complementation constitutes a prominent example [13]. In one of the...