Download PDF
Synlett 2013; 24(2): 150-156
DOI: 10.1055/s-0032-1317498
account
© Georg Thieme Verlag Stuttgart · New York

From Commercial Enzymes to Biocatalysts Designed by Protein Engineering

Uwe T. Bornscheuer*
Department of Biotechnology and Enzyme Catalysis, Institute of Biochemistry, Greifswald University, Felix-Hausdorff-Str. 4, 17487 Greifswald, Germany   Fax: +49(3834)86794367   Email: uwe.bornscheuer@uni-greifswald.de
› Author Affiliations
Further Information

Publication History

Received: 24 August 2012

Accepted after revision: 01 October 2012

Publication Date:
09 November 2012 (online)

 
 PDF Download Permissions and Reprints All articles of this category


Abstract

This account provides a personal view on the development of biocatalysis over the last two decades. Examples include the use of commercial enzymes, such as lipases, (recombinant) esterases, transaminases, and Baeyer–Villiger monooxygenases for the synthesis of optically pure compounds. The opportunity provided by modern protein engineering methods to tailor design an enzyme for a given scientific problem (substrate scope, selectivity, stability) is emphasized together with concepts to boost this technology in terms of timelines and success.

1 Introduction

2 Unexpected Discoveries

2.1 To Protect and Serve

2.2 ‘Abnormal’ Access to β-Amino Acids

3 Defined Enzyme Is Better Than Crude Extract

4 New Horizons Opened by Protein Engineering

4.1 Random Mutagenesis Can Give Random Results

5 Exploring Sequence and Structure Databases

5.1 Massive Alignment Identifies Evolutionary Variations

5.2 Fixing Wrong Annotations Can Yield a Toolbox of Novel Enzymes

6 Conclusion


#

Key words

asymmetric catalysis - biocatalysis - enantioselectivity - enzymes - kinetic resolution - protein engineering

Biographical Sketch

Zoom Image

Uwe T. Bornscheuer (born 1964) studied chemistry (1985–1990) at the University of Hannover (Germany) and completed his doctorate at the same university in 1993. He then held a postdoctoral position (1993–1994) at the University of Nagoya (Japan), where he worked on enzymatic lipid modification. In 1998, he completed his Habilitation at the University of Stuttgart (1994–1998), and he was appointed professor at Greifswald University in 1999. In 2003/2004, he spent a six-month sabbatical at Diversa Corp., San Diego (USA), and in 2010 he was at Kyoto University (Japan) for two months. Bornscheuer has written and edited several books. He is Editor-in-Chief of the European Journal of Lipid Science and Technology and Co-chairman of the journal ChemCatChem. In 2008, he received the Biocat2008 Award for his innovative work in biocatalysis and in 2012 he was awarded the Chevreul Medal for his pioneering work in enzymatic lipid research. His current research interest is protein engineering of enzymes from various classes for application in organic synthesis or in lipid modification.

1

Introduction

Recently, we suggested that modern biocatalysis developed in three waves.[ 1 ] The first wave commenced a century ago and was based on the discovery that enzymes, microorganism or cell extracts, could be used for chemical transformations. For instance, Rosenthaler used a plant extract for the synthesis of (R)-mandelonitrile from benzaldehyde and hydrogen cyanide in 1908.[ 2 ] However, until about 20–25 years ago, researchers who wished to use biocatalysts for organic synthesis had access to only a handful of commercially available enzymes for a given reaction.[ 3 ]

The second wave started in the 1980s and 1990s and was driven by early protein engineering techniques, which enabled the creation of enzymes for the synthesis of unusual synthetic intermediates. The focus was usually on the improved stability and selectivity of the enzyme. The current third wave exploits advanced molecular biology, bioinformatics, and high-throughput tools to address challenging problems in biocatalysis as summarized in our review: ‘In the past, an enzyme-based process was designed around the limitations of the enzyme; today, the enzyme is engineered to fit the process specifications’.[ 1 ]

When I started my Ph.D. in 1991 at Hannover University, elements of the first wave of biocatalysis were still in use. For instance, in the enzyme class of hydrolases (EC 3) – which I used during my studies – only about 10–15 lipases and pig liver esterase (crude extract) were commercially available to solve a given synthetic chemistry problem, such as the kinetic resolution of a racemic alcohol (Scheme [1]). Thus, one tested these few enzymes with respect to activity, conversion, and especially enantioselectivity under certain standard conditions (e.g., hydrolysis of the corresponding racemic acetate in phosphate buffer, pH 7, 50 mM). If this did not result in optically pure product in reasonable yield, then one alternative was to perform a transesterification of the racemic alcohol in organic solvents (Scheme [1]).

Zoom Image
Scheme 1 Hydrolysis versus transesterification in the kinetic resolution of a racemic acetate or alcohol by a lipase; note that the absolute configuration of the alcohol formed is opposite in the two approaches as the lipase maintains its enantiopreference

This was a very popular topic at that time following Zaks and Klibanov’s publication of their seminal discovery that porcine pancreatic lipase is highly active in organic solvents even at 100 °C.[ 4 ] The advantage of this ‘medium engineering’ was that different solvents could have a substantial – unfortunately unpredictable – effect not only on reaction rate, but also on the enantioselectivity of an enzyme. Furthermore, the acyl donor could be chosen in a flexible manner with respect to acyl chain length and type of activation. In particular, enol esters such as vinyl acetate became the standard reagents in hydrolase-catalyzed kinetic resolutions because the reaction equilibrium is shifted substantially by keto–enol tautomerization to the volatile acetaldehyde being formed as the byproduct.

Overall, one needed to test numerous solvents (including water-miscible vs water-immiscible ones) and immobilization methods to ensure the stability of the biocatalysts in addition to the variation of the acyl donor; hence perspiration not inspiration was often the key to success. Despite all these issues, numerous successful examples were published during that time on the use of lipases in organic synthesis[ 5 ] for the preparation of optically pure alcohols and carboxylic acids. This was not only important for academic research, but also for industrial-scale production as many of the compounds represented pharmaceutical intermediates or general building blocks.[ 6 ] As stated above, a process needed to be designed around the limitations of the enzyme to cope with stability, activity, solvent tolerance, pH, and temperature optima (see below for the current status using advanced protein engineering).

The same was true in the 1990s for other enzyme classes. For example, for the reduction of ketones to yield optically active chiral alcohols either baker’s yeast [notably producing various ketoreductases/alcohol dehydrogenases (ADHs) with different selectivities] or a few animal-derived (such as horse liver ADH) or microbial (such as Thermoanaerobium brockii ADH, Lactobacillus species ADH) enzymes were available.


# 2

Unexpected Discoveries

All scientific projects should start with a good plan, i.e. choice of enzyme, synthesis of substrates, target compounds, analytical methods, etc. If the budget comes from funding organizations, a grant must be well prepared with detailed work packages and milestones. I am quite sure that very often a scientist (Ph.D. student, postdoctoral scientist, etc.) looking back at an initial plan from three years ago realizes that in the meantime the project has developed into something significantly different. Dead ends, deviations, and loops are waiting around each corner of a project. The major challenge for the scientist is to find a way around these issues, and the task of the supervisor is – besides his support to find the best solution and ensure the budget – to be highly optimistic, motivating the student to believe that there is always light at the end of the tunnel. The examples from my research given below should demonstrate that it is often worth following different routes, reading as much literature as possible, and keeping one’s eyes open for unexpected things to happen.

2.1

To Protect and Serve

It is well documented in the literature that lipases usually show high enantioselectivity in the kinetic resolution of secondary alcohols if a small substituent, such as a methyl group, is on one side of the chiral center and a large substituent, such as a phenyl ring, is on the other. This led to the development of the so-called Kazlauskas rule.[ 7 ] Hence, organic chemists can predict – based on the substrate structure – which lipase would possibly be useful, which enantiomer should form preferentially, and whether chances were good for high selectivity (this of course would not replace experimental determination). The situation is different for primary alcohols bearing an additional methylene group between the chiral center and the hydroxy group. Efforts by Kazlauskas, reported in 1995, to transfer his rule to primary alcohols failed for compounds that have an oxygen atom attached to the stereocenter, and he stated that for these compounds the ‘rule’ was only as good as guessing.[ 8 ]

During that time, we wished to synthesize 2,4,6-trifunctionalized tetrahydropyrans by the desymmetrization of meso-precursors to provide access to optically pure 2,4,6-trifunctionalized C-glycosides (Scheme [2]).[ 9 ] Based on Kazlauskas’ observation, we expected low selectivity of lipases towards this primary alcohol, and indeed with the keto derivative at best 70% enantiomeric excess (ee) could be obtained. In contrast, protecting the free carbonyl group of the meso-diacetate as a benzyl ether or ketal allowed desymmetrization with a lipase in high chemical yield and excellent optical purity (Scheme [2]). Hence, variation of the substrate structure solved this problem, even if the result was unexpected.

Zoom Image
Scheme 2 Lipase-catalyzed hydrolysis of meso-precursors; the meso-keto substrate gave the monoacetate in only modest optical purity, but introduction of different protecting groups, such as in the ketal compound, resulted in excellent enantioselectivity

More recently, we made a similar observation in the kinetic resolution of 3-aminopyrrolidine (3AP) with commercial ω-transaminases,[ 10 ] enzymes belonging to the class of transferases (EC 2). Similar to the situation with lipases in the 1990s, only a few transaminases were commercially available. Again, the enzymes were investigated under various reaction conditions using 3AP and resulted in only modest reaction rate and reasonable optical purity.

As the product is a chiral compound required for subsequent chemical transformations, we also investigated a set of different N-protected (Bn, Boc, Cbz) derivatives. To our surprise, the introduction of a protecting group lead to substantial rate enhancement (up to 50-fold), and optical purity increased from 86 to 99% ee (at 50% conversion) in the case of the N-benzyloxycarbonyl-protected 3AP compared with the unprotected substrate (Scheme [3]). However, benzyl-protection of 3AP had no influence on enantioselectivity, only on reaction rate. As an explanation for this observation, we suggested the different flexibility of the benzyl- or benzyloxycarbonyl-protected 3AP as confirmed by NMR spectroscopy.[ 10 ] At this conclusion, one could be tempted to quote a professor at Hannover University who stated, ‘Give me your data and I will come up with a theory’.

Zoom Image
Scheme 3 The transaminase-catalyzed kinetic resolution of racemic 3-aminopyrrolidine, which surprisingly proceeded much faster and with higher enantioselectivity when the ring nitrogen was protected

# 2.2

'Abnormal' Access to β-Amino Acids

β-Amino acids are important compounds in the synthesis of a variety of drugs, but not many facile enzymatic routes to obtain them have been described.

We discovered that certain Baeyer–Villiger monooxygenases[ 11 ] (BVMOs) can convert aliphatic ketones into the corresponding esters, while the majority of these biocatalysts prefer cyclic (such as cyclohexanone) or aryl aliphatic ketones (such as 4-hydroxyacetophenone). The regioselectivity of these enzymatic Baeyer–Villiger oxidations using molecular oxygen and the cofactor NAD(P)H [nicotinamide adenine dinucleotide (phosphate)] is identical to that of the chemical reaction using peracids or hydrogen peroxide. In both cases, the ester or lactone obtained results from migration in the Criegee intermediate of the more highly substituted residue, which is more able to stabilize the developing positive charge.

Next, we investigated the kinetic resolution of 4-hydroxyalkan-2-ones and could thus provide access to optically pure diols.[ 12 ] Logically, we then verified other substituents at the β-position using a large set of BVMOs. Surprisingly, we observed that with certain enzymes, the insertion of oxygen occurs at the ‘abnormal’ position. Depending on the regioselectivity of the enzyme, we could obtain either 2-amino-1-hydroxy-substituted products – the ‘normal’ product – or β-amino acids with excellent selectivity as exemplified for β-leucine (Scheme [4]).

Zoom Image
Scheme 4 The kinetic resolution of β-amino-substituted ketones; depending on the regioselectivity of the BVMO used, the kinetic resolution yields (after subsequent lipase-catalyzed ester hydrolysis) the ‘normal’ chiral amino alcohol (left) or the ‘abnormal’ β-amino acid (right) (PG = protecting group)

We are currently trying to understand the molecular reasons for this unusual regioselectivity using protein engineering to create BVMO variants with defined regioselectivity.


#
# 3

Defined Enzyme Is Better Than Crude Extract

An organic chemist would hesitate to use a reagent or catalyst with only modest purity. In biocatalysis, it is common practice to use commercial crude preparations as the purification of a protein is costly and often results in less stable enzymes. This is acceptable as long as no undesired side reactions occur.

Pig liver esterase (PLE) was used for decades as a crude extract (obtained simply by acetone precipitation from homogenized pig liver tissue) in organic synthesis for two major reasons: (i) its preparation was very cheap and (ii) it is a biocatalyst with broad substrate scope and exceptionally high selectivity in desymmetrizations.[ 13 ] Unfortunately, it has been known since 1979 that PLE consists of several isoenzymes.[ 14 ] Moreover, it has been reported that the varying composition of isoenzymes from batch to batch can have profound effects on enantioselectivity.[ 15 ] This feature has substantially hampered the reliable application of crude PLE in industrial applications.

We therefore started the endeavor of identifying the nucleotide (and protein) sequences encoding these isoenzymes (starting with the extraction of mRNA isolated from fresh pig liver obtained from a butcher in a nearby store which was then transcribed to cDNA followed by gene identification) to enable the recombinant production of PLE. By this means we discovered five isoenzymes of PLE differing in 3–21 amino acid residues.[16] [17] The successful expression of active PLE isoenzymes at defined quality was first achieved in the yeast Pichia pastoris [ 18 ] and then later in E. coli.[ 19 ]

The application of these pure isoenzymes revealed that they indeed differ substantially in their enantioselectivity. For instance, the kinetic resolution of a series of acetates of secondary alcohols clearly showed that with two isoenzymes, the (S)-alcohol of 3 is formed in good to excellent ee, whereas other isoenzymes provide access to the pure R-enantiomer (Figure [1]). Not surprisingly, the commercial preparation from Fluka (crude pig liver extract) yields the almost racemic product as the various isoenzymes complement each other. Hence, an organic chemist using the crude preparation would have surely stated that PLE does not work for this substrate! A similar pattern of selectivity was observed for the isoenzymes in the desymmetrization of the meso-diacetate cis-3,5-diacetoxycyclopent-1-ene.[ 17 ] A further isoenzyme (named APLE) discovered by researchers from Graz in collaboration with the company DSM (The Netherlands) is already used by DSM on a large scale for the production of methyl (4E)-5-chloro-2-isopropylpent-4-enoate.[ 20 ]

Zoom Image
Figure 1 Substrates 1–4 used for kinetic resolution with pure isoenzymes of recombinant pig liver esterase (PLEs 1–5) or with commercial pig liver esterase (Fluka-PLE, crude extract from pig liver); the graph compares the enantioselectivity of the reactions and shows that with certain enzymes and acetates the (S)-alcohol is preferentially formed, whereas in other cases opposite enantiopreference occurs leading to the (R)-alcohol

# 4

New Horizons Opened by Protein Engineering

Although the discovery of recombinant DNA technology in the late 1970s was a major breakthrough to make protein production simpler, that still did not solve the issue of generating an enzyme in the laboratory with desired biocatalytic properties. Consequently, many attempts to establish biotransformations suffered from the lack of easy methods to alter enzyme properties. Of course, rational protein design based on X-ray structures of proteins was possible in the early 1990s, but modeling for the prediction of positions for mutagenesis was rather limited to bioinformatic experts and, in addition, computers and software were expensive.

During my time in Stuttgart at the Institute of Technical Biochemistry headed by Professor Rolf D. Schmid, I had the great opportunity to learn – trained as a chemist – micro- and molecular biology methods, as well as being introduced to bioinformatics by Jürgen Pleiss in Schmid’s group. Around the same time, a new concept in protein engineering, directed (molecular) evolution, was pioneered by Frances Arnold’s group[ 21 ] and Willem Stemmer.[ 22 ] In brief, these methods combine random mutagenesis of enzyme-encoding genes with screening or selection to find desired variants in mutant libraries. Consequently, one can target the improvement of an enzyme without the requirement of a protein structure! I was quickly fascinated by this technology, and my first success was the alteration of the substrate specificity of an esterase to accept a bulky compound not hydrolyzed by the wild-type enzyme.[ 23 ] The mutant library was created using a mutator E. coli strain. The selection of active variants was performed on agar plates and was based on a pH drop (owing to released acid after ester cleavage) and better growth of E. coli (when glycerol was released from the substrate).

4.1

Random Mutagenesis Can Give Random Results

However, in another project we learned the hard way that directed evolution by error-prone PCR (polymerase chain reaction) cannot solve all protein engineering problems when we tried to create an esterase with activity towards tertiary alcohols. Later, we learned from bioinformatic analysis that a distinct motif [‘GGG(A)X motif’, where G denotes glycine, A alanine, and X any amino acid] in the oxyanion hole in the active site region determines whether a lipase or esterase is active towards the sterically demanding esters of tertiary alcohols.[24] [25] This was a major breakthrough and we could identify many tertiary alcohol accepting hydrolases, which were subsequently improved with respect to enantioselectivity and substrate scope by protein engineering.

As we started from an esterase with only one glycine residue in this pocket (GX motif), the probability that we would be able to encode on a genetic level two new adjacent glycine residues at these specific locations by random mutagenesis was extremely low as this requires several simultaneous nucleotide exchanges owing to the codon usage. Without the bioinformatic analysis of the problem, my Ph.D. student would possibly still be screening agar plates for active clones.

In a project that started after I became a professor in Greifswald, we wished to create esterase mutants by protein engineering for the highly selective kinetic resolution of secondary alcohols for which – in agreement with the Kazlauskas rule – no lipase or esterase showed sufficient selectivity, as confirmed through an extensive screening.[ 26 ] One of these compounds was (S)-but-3-yn-2-ol, a versatile chiral building block. We chose an esterase from Pseudomonas fluorescens (PFE) as this enzyme could be easily functionally overexpressed in E. coli. The gene encoding PFE was subjected to error-prone PCR followed by the creation of a mutant library in microtiter plates (MTPs).

The key to successful directed evolution is a high-throughput assay, which is fast, reliable, and preferentially uses the true substrates (not a surrogate, i.e. an ester of a bulky chromophore). Previously, we described an assay suitable for screening in MTPs based on the release of acetic acid from an esterase-catalyzed hydrolysis. The acid is then converted in a three-enzyme cascade into an NADH signal stoichiometrically coupled to the acetic acid released.[ 27 ] After screening approximately 7000 esterase mutants, a variant was identified with good enantioselectivity (E = 89), but low activity.[ 28 ] Further analysis revealed that this variant contained three mutations (Ile76Val/Gly98Ala/Val175Ala), and, in contrast to the wild-type esterase, expression in E. coli resulted mostly in inclusion bodies (inactive protein aggregates).

The major question was then which mutation(s) contributed to improved selectivity and which were responsible for proper folding of the active enzyme? Structural analysis strongly suggested that Gly98Ala being close to the active site improved selectivity. To verify this, we constructed all three single and three double mutants and analyzed them in detail. Luckily, but unexpectedly from the modeling predictions, it turned out that the Gly98Ala mutation affected protein folding substantially and the double mutant Ile76Val/Val175Ala resulted in a highly selective variant (E = 96) with properties (expression, specific activity, temperature, and pH profile) similar to the wild-type esterase.[ 28 ]

The lesson we learned is that a sensitive assay is very important – otherwise we would have missed the weakly active triple mutant – and that structural analysis of the variants by computer modeling does not replace experiments in the laboratory.


#
# 5

Exploring Sequence and Structure Databases

The success of protein engineering is substantially facilitated by information deposited in public databases, the most important of which are protein sequence databases fostered mostly by genome sequencing and especially metagenome[ 29 ] projects. A major leap forward in this field has been the development of next generation sequencing techniques substantially increasing the genetic information in public databases, with currently more than 20 million protein sequences waiting to be explored. A range of software tools then allow retrieval, comparison, and analysis of this information. For instance, the tool BLAST (Basic Local Alignment Search Tool) quickly identifies proteins homologous to the enzyme under investigation. This alignment can then be used to discover motifs [see example above for the resolution of tertiary alcohols with esterases having the GGG(A)X motif] or new enzymes (which then are cloned, expressed, and analyzed) or to guide protein engineering experiments.

In parallel, structure-guided approaches have benefited from a rapid increase in protein structures deposited in the Brookhaven protein structure database (www.pdb.org). Over the last decade, the repository has grown approximately 4.5-fold to >77,000 protein structures. This facilitates both rational protein design as well as directed evolution approaches, as structural alignment of related proteins helps to identify distinct similarities and differences, guiding the more reliable design of mutant libraries.

5.1

Massive Alignment Identifies Evolutionary Variations

In a recent project, we took advantage of the huge amount of information stored in protein databases when we planned to increase the enantioselectivity of the esterase PFE. Instead of random mutagenesis of the entire protein, we identified four amino acid residues in the binding pocket which we assumed were responsible for the selectivity of the enzyme. One option was then to perform saturation mutagenesis separately at each position and try to identify the best hits.[30] [31] This approach however neglects synergistic and cooperative effects. The alternative was simultaneous saturation mutagenesis at all four positions, but this results in 160000 (204) possible combinations and requires screening of millions of clones.[ 31 ]

To reduce the library size, but maintain good chances of finding the desired amino acid substitutions, we used the bioinformatic tool 3DM. This commercial program creates a structure-based sequence alignment of all enzymes known from databases, in this case those within the α/β-hydrolase fold superfamily.[ 32 ] 3DM then allows one to filter this information and to identify only those amino acid substitutions that have occurred frequently during the natural evolution of all the esterases in the superfamily.

An alignment of >2800 related enzymes limited the possible amino acid substitutions to a few hundred variants – we dubbed them ‘small, but smart’ libraries – which were then created and investigated for improved enantioselectivity. This led to the discovery of a double and a triple mutant with excellent E values (E mutant = 80, E WT = 3.2).[ 33 ] The same concept was successfully applied to improve the thermostability of the esterase by targeting three residues at the surface of the enzyme.[ 34 ]

For both projects, it took only about three months to get the final results, which is pretty fast bearing in mind the required laborious molecular biology work, the high-throughput screening, and the biochemical characterization of best variants.


# 5.2

Fixing Wrong Annotations Can Yield a Toolbox of Novel Enzymes

As pointed out above, more than 20 million protein sequences can be found in databases. However, the annotation of a specific function of a newly discovered (but not biochemically characterized) protein sequence is done by comparison to known functions of related enzymes. Consequently, this can result in misleading or simply inaccurate annotations as small variations in a protein sequence can lead to a different function of the enzyme. This can be frustrating: if an enzyme has a wrongly predicted function from database analysis, a researcher may use the wrong substrates to verify its activity without any success.

We recently took advantage of these misleading annotations to identify novel sequences encoding (R)-selective amine transaminases (ATAs). These pyridoxal 5′-phosphate (PLP) dependent enzymes are useful as they enable an asymmetric synthesis of chiral amines from ketones. This distinguishes them from α-transaminases, which convert α-keto acids into α-amino acids.

Although (S)-selective ATAs were already described in the literature, no protein structure was available to guide rational design. As the target amines required the R-configuration, we envisaged classical screening of microorganisms or protein engineering either to convert an (S)-ATA into an (R)-ATA or to evolve an (S)-amino acid transaminase to accept substrates lacking the carboxyl group – turning it into an artificial ATA. This mutant would then be an (R)-ATA owing to a change in the priority of the substituents according to the Cahn–Ingold–Prelog rule. Unfortunately, in our hands all of these approaches were unsuccessful.

At that time, more than 5000 protein sequences of putative (or real) transaminases from a certain fold class of PLP-dependent enzymes were deposited in public databases. We thus developed an algorithm based on the identification of key amino acid motifs and residues that clearly distinguish (S)- from (R)-amino acid transaminase and that are responsible for harboring a carboxylic group in the active site. Applying this algorithm sorted out the majority of all the proteins, with 20 sequences remaining (corresponding to only 0.4% of all the sequences in the database). According to our theory, these 20 enzymes would have R-enantiopreference and convert ketones into chiral amines. We ordered all the encoding genes, produced and purified the enzymes, and were pleased to find that our in silico prediction of their function was perfectly correct.[ 35 ] Furthermore, we could show that these (R)-ATAs were synthetically useful by demonstrating the asymmetric synthesis of a set of 12 chiral amines using 7 of these enzymes.[ 36 ] It is also important to mention that the entire discovery and preliminary verification took just a few months.

One might compare the development of protein engineering over the last decades with improvements in weather forecasting: Initially based on decades of experience (drawn from adages by farmers or sailors), which represents the first wave, predictions became substantially better by having many observation stations, weather balloons, and dedicated meteorological models (second wave). Now sophisticated computer programs allow a rather reliable forecast for several days (third wave). In protein engineering, we are now close or even at that third stage.


#
# 6

Conclusion

Without any doubt, biocatalysis is a highly interesting and innovative field. Taking advantage of protein engineering tools opened many new horizons for me and significantly advanced finding solutions to the scientific problems studied in my group. It also provided the fun opportunity to work with a talented interdisciplinary team with experience in organic chemistry, micro- and molecular biology, bioinformatics, computer modeling, high-throughput screening, analytical methods, and other methodologies.


#
#

Acknowledgment

I am grateful to the many talented and encouraged group members who contributed to the various projects during my career. I also would like to thank my former supervisors for their inspiration and motivation.



   Permissions and Reprints All articles of this category
Zoom Image
Zoom Image
Scheme 1 Hydrolysis versus transesterification in the kinetic resolution of a racemic acetate or alcohol by a lipase; note that the absolute configuration of the alcohol formed is opposite in the two approaches as the lipase maintains its enantiopreference
Zoom Image
Scheme 2 Lipase-catalyzed hydrolysis of meso-precursors; the meso-keto substrate gave the monoacetate in only modest optical purity, but introduction of different protecting groups, such as in the ketal compound, resulted in excellent enantioselectivity
Zoom Image
Scheme 3 The transaminase-catalyzed kinetic resolution of racemic 3-aminopyrrolidine, which surprisingly proceeded much faster and with higher enantioselectivity when the ring nitrogen was protected
Zoom Image
Scheme 4 The kinetic resolution of β-amino-substituted ketones; depending on the regioselectivity of the BVMO used, the kinetic resolution yields (after subsequent lipase-catalyzed ester hydrolysis) the ‘normal’ chiral amino alcohol (left) or the ‘abnormal’ β-amino acid (right) (PG = protecting group)
Zoom Image
Figure 1 Substrates 1–4 used for kinetic resolution with pure isoenzymes of recombinant pig liver esterase (PLEs 1–5) or with commercial pig liver esterase (Fluka-PLE, crude extract from pig liver); the graph compares the enantioselectivity of the reactions and shows that with certain enzymes and acetates the (S)-alcohol is preferentially formed, whereas in other cases opposite enantiopreference occurs leading to the (R)-alcohol