Anna Joëlle Ruff

To get what we aim for – progress in diversity generation methods

Protein re‐engineering by directed evolution has become a standard approach for tailoring enzymes in many fields of science and industry. Advances in screening formats and screening systems are fueling progress and enabling novel directed evolution strategies, despite the fact that the quality of mutant libraries can still be improved significantly. Diversity generation strategies in directed enzyme evolution comprise three options: (a) focused mutagenesis (selected residues are randomized); (b) random mutagenesis (mutations are randomly introduced over the whole gene); and (c) gene recombination (stretches of genes are mixed to chimeras in a random or rational manner). Either format has both advantages and limitations depending on the targeted enzyme and property. The quality of diverse mutant libraries plays a key role in finding improved mutants. In this review, we summarize methodological advancements and novel concepts (since 2009) in diversity generation for all three formats. Advancements are discussed with respect to the state of the art in diversity generation and high‐throughput screening capabilities, as well as robustness and simplicity in use. Furthermore, limitations and remaining challenges are emphasized ‘to get what we aim for’ through ‘optimal diversity’ generation.

Lessons from diversity of directed evolution experiments by an analysis of 3,000 mutations

Diversity generation by random mutagenesis is often the first key step in directed evolution experiments and screening of 1,000–2,000 clones is in most directed evolution campaigns sufficient to identify improved variants. For experimentalists important questions such as how many positions are mutated in the targeted gene and what amino acid substitutions can be expected after screening of 1,000–2,000 clones are surprisingly not answered by a statistical analysis of mutant libraries. Therefore three random mutagenesis experiments (epPCR with a low‐ and a high‐mutation frequency and a transversion‐enriched sequence saturation mutagenesis method named SeSaM‐Tv P/P) were performed on the lipase BSLA and in total 3,000 mutations were analyzed to determine the diversity in random mutagenesis libraries employed in directed evolution experiments. The active fraction of the population ranged from 15% (epPCR‐high), to 52% (SeSaM‐Tv P/P), and 55% (epPCR‐low) which correlates well with the average number of amino acid substitutions per protein (4.1, 1.6 and 1.1). In the epPCR libraries transitions were the predominant mutations (>72%), and >82% of all mutations occurred at A‐ or T‐nts. Consecutive nucleotide (nt) mutations were obtained only with a low fraction (2.8%) under highly error‐prone conditions. SeSaM‐Tv P/P was enriched in transversions (43%; >1.7‐fold more than epPCR libraries), and consecutive nt mutations (30.5%; 11‐fold more than epPCR‐high). A high fraction of wild‐type BSLA protein (33%) was found in the epPCR‐low mutant library compared to 2% in epPCR‐high and 13% in SeSaM‐Tv P/P. An average of 1.8–1.9 amino acid substitutions per residue was obtained with epPCR‐low and ‐high compared to 2.1 via SeSaM‐Tv P/P. The chemical composition of the amino acid substitutions differed, however, significantly from the two epPCR methods to SeSaM‐Tv P/P. Biotechnol. Bioeng. 2014;111: 2380–2389.

Directed Evolution of P 450 BM 3 into a p‐Xylene Hydroxylase

Regioselective hydroxylation of aromatic rings is a “dream reaction” in organic chemistry and is of high interest for the production of precursors for pharmaceuticals and flavors. In particular, the demand for phenol and alkylphenols as a feedstock for resins, plastics, and bisphenol A has increased constantly throughout the last century. Phenols are produced typically through the three-step cumene process, also known as the “Hock Process”. Alternatively, they can be isolated from natural sources such as coal tar, biomass, or by gasification processes. The latter often require intensive extraction procedures with high energy consumption, especially for the separation of isomeric alkylphenols. The “Hock Process” is highly profitable, in spite of its low product yields ( 5 % of the initially used benzene is converted to phenol) ; this owes mainly to the valuable acetone byproduct. Novel synthetic routes to phenols focus mainly on the development of metal catalysts, which generate reactive oxygen species for the hydroxylation of benzene educts. Direct hydroxylation with enzymes can be an attractive alternative; it reduces reaction and purification steps, as well as waste generation and process energy demands. Protein engineering methods such as semi-rational design and directed evolution allow these biocatalysts to be tailored to cost-effective production conditions. Cytochrome P 450 monooxygenases (CYPs), a protein superfamily that contains more than 5000 cloned members, catalyze oxygenation reactions with molecular oxygen at room temperature. Direct aromatic ring hydroxylation is a synthetically attractive reaction that has been reported for several P 450 monooxygenases. Whitehouse et al. recently described the rational driven design of the P 450 monooxygenase BM 3 (CYP 102 A 1) towards regioselective aromatic hydroxylation of substituted benzenes. A reaction mechanism for understanding aromatic hydroxylation of oand m-xylene has been proposed, however p-xylene was not investigated as a substrate. The p-xylene isomer was reported to be exclusively hydroxylated at the a-position. 10] Only a few enzymes (toluene 4-monooxygenases and microsomal P 450 monooxygenases) were reported to catalyze the aromatic hydroxylation of p-xylene to the corresponding phenol. 12] Unfortunately, enzymes such as toluene 4-monooxygenase suffer from low activity (kcat = 36 min ), stability (four-protein complex), and selectivity (20 % aromatic and 80 % benzylic hydroxylation) with p-xylene as the substrate. Nonetheless, direct aromatic hydroxylation of p-xylene to 2,5-dimethylphenol (2,5-DMP) is of high interest for the production of temperature stable polymers (>500 K), as is 2,5-DMP itself, which can act as a building block for “next-generation” plastics. We describe the semi-rational engineering of the self-sufficient monooxygenase P 450 BM 3 towards the direct aromatic hydroxylation of p-xylene. To achieve this goal, we selected and iteratively saturated five amino acid residues—Arg 47 (R 47), Tyr 51 (Y 51), Phe 87 (F 87), Ala 330 (A 330), and Ile 401 (I 401)—that had been reported to influence either substrate selectivity, activity, or both (Figure 2). 9, 14, 15] Besides using a sensitive and regioselective screening assay for phenols, we confirmed all products from P 450 BM 3 hydroxylation by GC (see Figure 1 and Experimental Section). Unexpectedly, the wild-type P 450 BM 3 showed a low activity for p-xylene (kcat = 68 min ; 45 % coupling efficiency), producing 2,5-DMP with high selectivity (>98 %), which contradicts a previous report. The latter finding can be attributed to higher substrate concentrations (1–30 mm p-xylene) used in the reaction setup. The recently proposed mechanism for oand m-xylene hydroxylation, however, does not exclude aromatic hydroxylation of p-xylene. In a first round of P 450 BM 3 mutagenesis, positions R 47/Y 51 were saturated simultaneously because both are key residues in controlling substrate access to the binding pocket (Figure 2). We obtained a variant M 1 (R 47 S/Y 51 W) with a more than seven-fold increase in product formation (kcat = 500 min 1 2,5-DMP) compared with the wild-type P 450 BM 3 (Table 1). Besides the increase in activity, variant M 1 showed a remarkable catalytic performance: the coupling efficiency increased by 20 % (total coupling 54 %). This was accompanied by a slightly decreased KM (from 7.9 to 7.1 mm). The bulky tryptophan residue on position 51 is thought to promote aromat-

Whole-cell double oxidation of n-heptane.

Biocascades allow one-pot synthesis of chemical building blocks omitting purification of reaction intermediates and expenses for downstream processing. Here we show the first whole cell double oxidation of n-heptane to produce chiral alcohols and heptanones. The concept of an artificial operon for co-expression of a monooxygenase from Bacillus megaterium (P450 BM3) and an alcohol dehydrogenase (RE-ADH) from Rhodococcus erythropolis is reported and compared to the widely used two-plasmid or Duet-vector expression systems. Both catalysts are co-expressed on a polycistronic constructs (single mRNA) that reduces recombinant DNA content and metabolic burden for the host cell, therefore increasing growth rate and expression level. Using the artificial operon system, the expression of P450 BM3 reached 81mgg(-1) cell dry weight. In addition, in situ cofactor regeneration through the P450 BM3/RE-ADH couple was enhanced by coupling to glucose oxidation by E. coli. Under optimized reaction conditions the artificial operon system displayed a product formation of 656mgL(-1) (5.7mM) of reaction products (heptanols+heptanones), which is 3-fold higher than the previously reported values for an in vitro oxidation cascade. In conjunction with the high product concentrations it was possible to obtain ee values of >99% for (S)-3-heptanol. Coexpression of a third alcohol dehydrogenase from Lactobacillus brevis (Lb-ADH) in the same host yielded complete oxidation of all heptanol isomers. Introduction of a second ADH enabled further to utilize both cofactors in the host cell (NADH and NADPH) which illustrates the simplicity and modular character of the whole cell oxidation concept employing an artificial operon system.

P-LinK: A method for generating multicomponent cytochrome P450 fusions with variable linker length.

Fusion protein construction is a widely employed biochemical technique, especially when it comes to multi-component enzymes such as cytochrome P450s. Here we describe a novel method for generating fusion proteins with variable linker lengths, protein fusion with variable linker insertion (P-LinK), which was validated by fusing P450cin monooxygenase (CinA) to the flavodoxin shuttle protein (CinC). CinC was fused to the C terminus of CinA through a series of 16 amino acid linkers of different lengths in a single experiment employing 3 PCR amplifications. Screening for 2-β-hydroxy-1,8-cineole production by CinA-CinC fusion proteins revealed that enzymatically active variants possessed linker lengths of more than 5 amino acids, reaching optimum enzyme activity at a linker length of 10 amino acids. Our P-LinK method not only minimizes experimental effort and significantly reduces time demands but also requires only a single cloning and transformation step in order to generate multiple linker variants (1 to 16 amino acids long), making the approach technically simple and robust.

A High-Throughput Screening Method to Reengineer DNA Polymerases for Random Mutagenesis

A screening system for directed evolution of DNA polymerases employing a fluorescent Scorpion probe as a reporter has been developed. The screening system has been validated in a directed evolution experiment of a distributive polymerase from the Y-polymerase family (Dpo4 from Sulfolobus solfataricus) which was improved in elongation efficiency of consecutive mismatches. The engineering campaign yielded improved Dpo4 polymerase variants one of which was successfully benchmarked in a sequence saturation mutagenesis experiment especially with regard to the desirable consecutive transversion mutations (>2.5-fold increase in frequency relative to a reference library prepared with Dpo4 WT). The Scorpion probe screening system enables to reengineer polymerases with low processivity and fidelity, and no secondary activities (i.e. exonuclease activity or strand displacement activity) to match demands in diversity generation for directed protein evolution.

Casting epPCR (cepPCR): A simple random mutagenesis method to generate high quality mutant libraries

During the last decade, directed evolution has become a standard protein engineering strategy to reengineer proteins for industrial applications under high stress conditions (e.g., high temperature, extreme pH, ionic liquids, or organic solvents). The most commonly employed method for diversity generation to improve biocatalysts for these properties is random mutagenesis by error‐prone polymerase chain reaction (epPCR). However, recent reports show that epPCR often fails to produce >70% of beneficial positions/amino acid exchanges which improve enzyme properties such as organic solvent or ionic liquid resistance. In this report, bsla (543 bp, small lipase gene from Bacillus subtilis) was divided into three fragments (147, 192, 204 bp). Each fragment was subjected to an epPCR with a high mutation load (22, 31, and 33 mutations per kb) in order to increase the number of identified beneficial positions while maintaining a fraction of active population which can efficiently be screened in agar plate or microtiter plate format. The use of this “casting epPCR” process termed as (cepPCR), doubles the number of identified beneficial positions (from 14% to 29%), when compared to standard epPCR for the BSLA enzyme model. A further increase to 39% of beneficial positions is obtainable through combination of cepPCR with the transversion biased sequence saturation mutagenesis (SeSaM) method. Furthermore, sequencing of up to 600 mutations per fragment provided valuable insights into the correlation of total throughput and number of identified beneficial positions as well as how an efficient balance of screening efforts to obtainable results can be achieved in directed evolution campaigns. Biotechnol. Bioeng. 2017;114: 1921–1927.

MuteinDB: the mutein database linking substrates, products and enzymatic reactions directly with genetic variants of enzymes

Mutational events as well as the selection of the optimal variant are essential steps in the evolution of living organisms. The same principle is used in laboratory to extend the natural biodiversity to obtain better catalysts for applications in biomanufacturing or for improved biopharmaceuticals. Furthermore, single mutation in genes of drug-metabolizing enzymes can also result in dramatic changes in pharmacokinetics. These changes are a major cause of patient-specific drug responses and are, therefore, the molecular basis for personalized medicine. MuteinDB systematically links laboratory-generated enzyme variants (muteins) and natural isoforms with their biochemical properties including kinetic data of catalyzed reactions. Detailed information about kinetic characteristics of muteins is available in a systematic way and searchable for known mutations and catalyzed reactions as well as their substrates and known products. MuteinDB is broadly applicable to any known protein and their variants and makes mutagenesis and biochemical data searchable and comparable in a simple and easy-to-use manner. For the import of new mutein data, a simple, standardized, spreadsheet-based data format has been defined. To demonstrate the broad applicability of the MuteinDB, first data sets have been incorporated for selected cytochrome P450 enzymes as well as for nitrilases and peroxidases. Database URL: http://www.MuteinDB.org

A whole cell biocatalyst for double oxidation of cyclooctane

A novel whole cell cascade for double oxidation of cyclooctane to cyclooctanone was developed. The one-pot oxidation cascade requires only a minimum of reaction components: resting E. coli cells in aqueous buffered medium (=catalyst), the target substrate and oxygen as environmental friendly oxidant. Conversion of cyclooctane was catalysed with high efficiency (50% yield) and excellent selectivity (>94%) to cyclooctanone. The reported oxidation cascade represents a novel whole cell system for double oxidation of non-activated alkanes including an integrated cofactor regeneration. Notably, two alcohol dehydrogenases from Lactobacillus brevis and from Rhodococcus erythropolis with opposite cofactor selectivities and one monooxygenase P450 BM3 were produced in a coexpression system in one single host. The system represents the most efficient route with a TTN of up to 24363 being a promising process in terms of sustainability as well.

Directed evolution of P450cin for mediated electron transfer

Directed evolution is a powerful method to optimize enzyme properties for application demands. Interesting targets are P450 monooxygenases which catalyze the stereo- and regiospecific hydroxylation of chemically inert C–H bonds. Synthesis employing P450s under cell-free reaction conditions is limited by low total turnover numbers, enzyme instability, low product yields and the requirement of the expensive co-factor NADPH. Bioelectrocatalysis is an alternative to replace NADPH in cell-free P450-catalyzed reactions. However, natural enzymes are often not suitable for using non-natural electron delivery systems. Here we report the directed evolution of a previously engineered P450 CinA-10aa-CinC fusion protein (named P450cin-ADD-CinC) to use zinc/cobalt(III)sepulchrate as electron delivery system for an increased hydroxylation activity of 1,8-cineole. Two rounds of Sequence Saturation Mutagenesis (SeSaM) each followed by one round of multiple site-saturation mutagenesis of the P450 CinA-10aa-CinC fusion protein generated a variant (Gln385His, Val386Ser, Thr77Asn, Leu88Arg; named KB8) with a 3.8-fold increase in catalytic efficiency (28 µM−1 min−1) compared to P450cin-ADD-CinC (7 µM−1 min−1). Furthermore, variant KB8 exhibited a 1.5-fold higher product formation (500 µM µM−1 P450) compared to the equimolar mixture of CinA, CinC and Fpr using NADPH as co-factor (315 µM µM−1 P450). In addition, electrochemical experiments with the electron delivery system platinum/cobalt(III)sepulchrate showed that the KB8 variant had a 4-fold higher product formation rate (0.16 nmol (nmol) P450−1 min−1 cm−2) than the P450cin-ADD-CinC (0.04 nmol (nmol) P450−1 min−1 cm−2). In summary, the current work shows prospects of using directed evolution to generate P450 enzymes suitable for use with alternative electron delivery systems.