Enzyme- and Transition Metal-Catalyzed Asymmetric Transformations

Application of Enzymatic (D)KR in Enantioselective Synthesis

Richard Lihammar © Richard Lihammar, Stockholm 2014 Cover picture: The Power of DKR!

ISBN 978-91-7649-008-2

Printed in Sweden by US-AB, Stockholm 2014 Distributor: Department of Organic Chemistry, Stockholm University ii Nothing shocks me. I’m a scientist.

- Indiana Jones

iii iv Abstract

Dynamic kinetic resolution (DKR) is a powerful method for obtaining compounds with high optical purity. The process relies on the combination of a kinetic resolution with an in situ racemization. In this thesis, a combination of an immobilized hydrolase and a transition metal-based racemization catalyst was employed in DKR to transform racemic alcohols and amines into enantioenriched esters and amides, respectively. In the first part the DKR of 1,2-amino alcohols with different rings sizes and N-protecting groups is described. We showed that the immobilization method used to support the lipase strongly influenced the stereoselectivity of the reaction. The second part deals with the DKR of C3-functionalized cyclic allylic alcohols affording the corresponding allylic esters in high yields and high ees. The protocol was also extended to include carbohydrate derivatives, leading to inversion of a hydroxyl substituted chiral center on the carbohydrate. The third part focuses on an improved method for obtaining benzylic primary amines. By using a novel, recyclable catalyst composed of Pd nanoparticles on amino-functionalized mesocellular foam, DKR could be performed at 50 °C. Moreover, Lipase PS was for the first time employed in the DKR of amines. In the fourth part DKR was applied in the total synthesis of Duloxetine, a compound used in the treatment of major depressive disorder. By performing a six-step synthesis, utilizing DKR in the enantiodetermining step, Duloxe- tine could be isolated in an overall yield of 37% and an ee >96%. In the final part we investigated how the enantioselectivty of reactions catalyzed by Candida Antarctica lipase B for δ-substituted alkan-2-ols are influenced by water. The results showed that the enzyme displays much higher enantioselectivity in water than in anhydrous toluene. The effect was rationalized by the creation of a water mediated hydrogen bond in the active site that helps the enzyme form enantiodiscriminating binding modes.

v vi List of Publications

This thesis is based on the following papers, which will be referred to by Roman numerals I-VI. Reprints were made with the kind permission from the publishers and the contribution by the author to each publication is clari- fied in Appendix A.

I. An Efficient Dynamic Kinetic Resolution of N-Heterocyclic 1,2- Amino Alcohols Richard Lihammar, Renaud Millet, Jan-Erling Bäckvall Advanced Synthesis & Catalysis 2011, 353, 2321–2327.

II. Enzyme- and Ruthenium-Catalyzed Dynamic Kinetic Resolution of Functionalized Cyclic Allylic Alcohols Richard Lihammar, Renaud Millet, Jan-Erling Bäckvall Journal of Organic Chemistry 2013, 78, 12114–12120.

III. Epimerization of Glycal Derivatives, by a Cyclopentadienyl- ruthenium Catalyst: Application to Metalloenzymatic DYKAT Richard Lihammar, Jerk Rönnols, Göran Widmalm, Jan-Erling Bäckvall Chemistry – A European Journal 2014, 20, 14756–14762.

IV. Chemoenzymatic Dynamic Kinetic Resolution of Primary Amines Using a Recyclable Palladium Nanoparticle Catalyst Together with Lipases Karl. P. J Gustafson, Richard Lihammar, Oscar Verho, Karin Engström, Jan-Erling Bäckvall Journal of Organic Chemistry 2014, 79, 3747–3751.

V. A Chemoenzymatic Dynamic Kinetic Resolution Approach to Enantiomerically Pure (R)- and (S)-Duloxetine Annika M. Träff, Richard Lihammar, Jan-Erling Bäckvall Journal of Organic Chemistry 2011, 76, 3917–3921.

vii VI. Investigation of the Impact of Water on the Enantioselectivity Displayed by CALB in the Kinetic Resolution of δ-Functionalized Alkan-2-ol Derivatives Bin Yang, Richard Lihammar, Jan-Erling Bäckvall Chemistry – A European Journal 2014, 20, 13517-13521.

viii Contents

1. Introduction ...... 1 1.1 Chirality ...... 1 1.2 Enzymatic catalysis in organic synthesis ...... 4 1.2.1 Hydrolases as enantioselective catalysts ...... 6 1.3 Enzymatic kinetic resolution...... 7 1.4 Dynamic kinetic resolution and racemization ...... 8 1.4.1 Shvo’s catalyst ...... 10 1.4.2 Racemization catalyst 5 ...... 11 1.5 General concerns regarding DKR ...... 13 1.6 Objectives of this thesis ...... 13

2. Dynamic Kinetic Resolution of Cyclic 1,2-Amino Alcohols ...... 15 2.1 Introduction...... 15 2.2 Results and discussion ...... 16 2.2.1 Optimization of KR and DKR of 11a and 13a ...... 16 2.2.2 Substrate scope ...... 18 2.2.3 Dynamic Kinetic Resolution ...... 20 2.3 Conclusion ...... 20

3. DKR and DYKAT of Functionalized Cyclic Allylic Alcohols. Application to Glycal Derivatives ...... 22 3.1 Introduction...... 22 3.2 Results ...... 23 3.2.1 Kinetic resolution ...... 23 3.2.2 Optimization of the DKR ...... 24 3.2.3 Scope of the reaction ...... 26 3.2.4 Synthetic application of 16g ...... 28 3.3 DYKAT on carbohydrate derivatives ...... 29 3.3.1 Background ...... 29 3.3.2 Epimerization and DYKAT of compound 21 and 22 ...... 30 3.3.3 Epimerization and DYKAT of compound 23 ...... 35 3.4 Conclusion ...... 36

ix 4. Dynamic Kinetic Resolution of Primary Amines, using a Recyclable Nanopalladium Catalyst for Racemization ...... 37 4.1 Introduction...... 37 4.2 Result and discussion ...... 40 4.2.1 Optimization of DKR ...... 40 4.2.2 Scope of the reaction ...... 42 4.2.3 Recycling study ...... 42 4.3 Conclusion ...... 44

5. Enantioselective Synthesis of Duloxetine ...... 45 5.1 Introduction...... 45 5.2 Result and discussion ...... 48 5.2.1 Kinetic resolution ...... 48 5.2.2 Racemization ...... 49 5.2.3 DKR ...... 50 5.2.4 Total synthesis of Duloxetine ...... 52 5.3 Conclusion ...... 54

6. Investigation of the Impact of Water on the Enantioselectivity Displayed by CALB in the Kinetic Resolution of δ-Functionalized Alkan-2-ol Derivatives ...... 55 6.1 Introduction...... 55 6.2 Results ...... 59 6.2.1 Transesterification vs. hydrolysis ...... 59 6.2.2 CALB A281S ...... 61 6.3 Discussion and Conclusion ...... 63

7. General conclusions and outlook ...... 64

Appendix A ...... 65

Appendix B ...... 66

Acknowledgements ...... 67

Reference ...... 70

x Abbreviations

Abbreviations and acronyms are used in agreement with standards of the subject.1 Only nonstandard and unconventional ones that appear in the thesis are listed here.

Ala Alanine AmP Aminopropyl Asp Aspartic acid CALB Candida Antarctica lipase B cat. catalyst CIP Cahn-Ingold-Prelog conv. Conversion DKR Dynamic kinetic resolution DYKAT Dynamic kinetic asymmetric transformation E Enantiomeric ratio ee Enantiomeric excess EWG Electron withdrawing group FDA Food and drug administration Glu Glutamic acid His Histidine KAT Kinetic asymmetric transformation KR Kinetic resolution MCF Mesocellular foam MS 4Å Molecular sieves, 4 Ångström nd. Not determined nr. No reaction o.n. Over night rac. Racemic sec. Secondary Ser Serine STS Surfactant-treated Subtilisin Carlsberg Thr Threonine

1. Introduction

1.1 Chirality An object is said to be chiral if it is not superimposable on its mirror image. A commonly used example to illustrate this phenomenon is the relationship between the right and the left hand. They both have the same kind of fingers connected to the palm; however, they are mirror images of each other and thus are not equal. The phenomenon of chirality is also found in chemistry, with chemical compounds having identical connectivity but different spatial arrangement. In organic chemistry, the asymmetry or chirality most often occurs on carbons that covalently binds to four different substituents. The substituents can be positioned in two different spatial arrangements, which lead to two non-superimposable molecules (Figure 1).

Figure 1. The two enantiomers of a carbon atom that covalently binds four different substituents.

The atom where the chirality arises is called a chiral center and, apart from at a carbon atom, chiral centers can also arise at heteroatoms such as sulfur or phosphorous.2 Organic molecules with identical connectivity but different spatial arrangement are called stereoisomers and the special case of molecules being mirror images of each other are called enantiomers. Enantiomers are classified as either (R) or (S) depending on the orientation of the substituents, according to the so called Cahn-Ingold-Prelog (CIP) priority rule.3 A mixture consisting of equal amounts of both enantiomers is called a racemate. If a molecule has two or more (n) chiral centers, the molecule can exist in three to 2n stereoisomeric forms. The forms that are not enantiomers of each other are called diastereomers (Figure 2).

Figure 2. Enantiomers and diastereomers of Chloramphenicol

The isomers that are enantiomers of one another will always display identical physical and chemical properties (e.g. boiling and melting points) in an achiral environment, with the exception of their ability to rotate plane- polarized light in different directions. However, in a chiral environment such as all biological systems, the enantiomers may differ in their interaction.4 As a consequence, chirality is an important property to take into consideration when investigating the interaction between organic molecules and biological receptors, enzymes and transporters. Often one of the enantiomers of an organic molecule shows higher biological activity than the other, and some- times it is only one that possesses a specific activity. In this case the other enantiomer might be inactive or, have effects that are harmful or even fatal.5 Two examples for which the enantiomeric pairs have completely different pharmacological responses are Chloramphenicol and Thalidomide (Figure 3). For Chloramphenicol the (R,R)-enantiomer possesses antibacterial properties while its (S,S)-enantiomer is inactive,6 and for Thalidomide the (R)- form is an effective sedative, whereas the (S)-form, is teratogenic and causes birth defects on fetuses.

Figure 3. Chiral molecules with enantiomers displaying different biological activity.

In 1992, as a consequence of the potential negative effects of certain enantiomers, the U.S. FDA issued a demand for the analysis of the pharmacological effects of both enantiomers of a racemic drug.7 This demand initiated an even greater intensity in both academia and industry to develop new methods for preparing compounds as single enantiomers. Moreover, pharmaceuticals started to be released as single enantiomer drugs. Some of the racemic drugs on the market were also re-issued as single enantiomers, which led to formu- lations possessing higher pharmacological activities. An example of this is racemic Omeprazole, sold as Losec or Prilosec, and its enantiopure counter- part Esomeprazole, sold as Nexium. (Figure 4).8

Figure 4. Omeprazole and Esomepraszole.

The methods available for obtaining enantiomerically enriched compounds can be divided into three categories: (i) resolution, where the two enantiomers of a racemic substrate mixture are separated from each other utilizing crystallization or a resolving agent, (ii) asymmetric synthesis, where a new chiral center is created from a prochiral substrate by induction from a second source of chirality and (iii) the chiral pool approach, using naturally occurring enantiomerically pure compounds such as amino acids, carbohydrates or terpenes as starting material for the synthesis of more complex molecules.9

3 1.2 Enzymatic catalysis in organic synthesis Enzymes are biomolecules that catalyze the formation and cleavage of chemical bonds. Even though enzymatic processes have been utilized by humans since ancient times for conservation and production of food (e.g. cheese making or brewing), it was first in the 19th century that the nature of enzymes started to be understood.10 A great breakthrough was made in 1897 when Buchner proposed that enzymes do not require the environment of a living cell to be active.11 This finding opened up for enzymes to be used in various areas such as production of food, leather, textiles, paper, and as bio- catalysts in chemical production. Because of enzymes being inexpensive, environmentally friendly catalysts reacting with high regio- and stereoselectivity they have been frequently used in chemical applications during the past century. One enzyme class that has attracted a lot of attention is hydrolases; catalyzing reactions where a chemical bond is cleaved by water.12 Nature uses these hydrolytic reactions to digest nutrients into smaller entities that can be utilized by the cells. For example, proteases break down proteins into smaller peptides or amino acids, and hydrolases hydrolyze lipids to glycerol and fatty acids.13 It was long believed that enzymes only worked in an aqueous environment and required water as a solvent to function.14 However in the early 80’s the scientific community started wondering, ”How much water do enzymes really need?”. When subjecting enzymes to reactions in almost completely dry organic solvents catalytic activities was indeed found for some enzymes, especially hydrolases such as proteases and lipases.15 It was later found that the water bound to the enzyme or a monolayer of water around the enzyme was sufficient to keep its flexibility and thus maintain its catalytic activity.16 There were several benefits of being able to perform the reaction in organic solvent: (i) better solubility of the substrate (usually organic molecules experience low solubility in aqueous solutions), (ii) simpler product recov- ery, (iii) modified enzyme properties and (iv) reversal of hydrolytic reactions. The latter feature makes it possible for hydrolases to catalyze the formation of an ester or amide from an alcohol or amine, respectively. For alcohols the reaction is denoted as an esterification if the acyl moiety origi- nates from an acid and a transesterification if the acyl donor is an ester. The mechanism of esterification and transesterification follows a ping- pong bi-bi mechanism, depicted for a sec-alcohol in Scheme 1.17 The reaction takes place in the active site of the enzyme where a serine is made nucleophilic by hydrogen bonds formed with nearby histidine and aspartic or glutamic acid.18 The serine attacks the acyl donor forming a tetrahedral intermediate (TI) that releases the first product (water or an alcohol) and forms an acyl-enzyme intermediate. This intermediate is subsequently attacked by the sec-alcohol forming a new TI. The negative charge of both the TIs is stabilized by hydrogen bonds in an oxyanion hole. Next, the TI breaks down,

4 releasing the free enzyme and an ester of a sec-alcohol. All steps are reversi- ble and hydrolysis of esters passes through the same TIs; however, the selectivity in the different reaction directions might differ due to solvent effects.13

Scheme 1. Transesterification mechanism of a serine hydrolase. Reaction starts in the upper left corner.

The reversibility gives rise to an important point to consider when designing a transesterification. Since the alcohol released from the acyl donor might function as a nucleophile hydrolyzing the formed ester it is desirable to choose an acyl donor that posses an alcohol moiety that is slow or incapable of acting as nucleophile. By selecting an appropriate acyl donor, the reaction can in principle be made irreversible (Scheme 2). Additionally, the type of leaving group and the length of the acyl chain can influence the performance of the enzyme and as a result different types of acyl donors have been studied and used in enzyme-catalyzed transesterifications (exemplified in Scheme 2 by isopropenyl acetate, 1, p-chlorophenyl acetate, 2, and 2,2,2- trifluoroethyl butyrate, 3).13

Scheme 2. Examples of acyl donors used in irreversible, enzyme catalyzed, transesterification reactions.

1.2.1 Hydrolases as enantioselective catalysts Since a hydrolase and its active site are chiral, the reactivity of the two enantiomers of a chiral substrate, such as a sec-alcohol, is different. This leads to the enzyme catalyzing the formation and hydrolysis of ester enantiomers at different rates. Since the ground states of the free enantiomers are equal, the level of stereoselectivity can be considered to be the energy difference between the two diastereomeric transition states formed during the catalysis of the (R) and (S) enantiomers. The TS is close in energy to the substrate bound TI, hence the selectivity difference can be approximated to the different relative energy of the two tetrahedral intermediates.19 The stereopreference for a lipase-catalyzed tranesterification of sec- alcohols can be predicted by the empirical rule proposed by Kazlauskas.20 The rule states that the size of the substituents residing at the chiral center determines which enantiomer that will be preferred. Assuming that the large- sized group has a higher priority than the medium-sized according to the CIP rule, lipases are (R)-stereoselective. The transesterification catalyzed by proteases from the families Subtilisin and Chemotrypsin works with a highly similar mechanism; however, their catalytic machinery is mirror imaged to that of lipases, hence they usually display (S)-stereoselectivity.10

Figure 5. The enantiomer preferred in a lipase- or a protease catalyzed reaction.

6 1.3 Enzymatic kinetic resolution The ability of lipases and proteases to enantioselectively catalyze transterification and hydrolysis reactions is exploited in the field of kinetic resolution (KR). In an enzymatic KR a racemic substrate is subjected to a biocatalyst that enantioselectively converts one substrate enantiomer into product, after which the product and the substrate can be separated from each other by standard purification methods (Scheme 3).

Scheme 3. Example of lipase catalyzed KR of a sec-alcohol.

In the ideal case, the enzyme is fully selective for one enantiomer leading to that a racemic substrate mixture is perfectly resolved into two compounds, both having an enantiomeric excess (ee) close to 100%. However, this is rarely the case, since the enzyme often catalyzes the conversion of the other enantiomer as well, but at a slower rate. Therefore, the resolution must be stopped before the conversion reaches 50%, in order to attain a high ee of the product. If instead the substrate is desired in a high ee, the reaction is required to proceed beyond 50% conversion. In both cases, a separation must be performed once the reaction is complete in order to access the resolved optically active compound. In order to quantify the enantioselectivity of an enzyme-catalyzed reaction, the dimensionless parameter ‘Enantiomeric Ratio’ (E value) has been developed. The E value expresses the ratio of the rate constants of the two enantiomers kfast/kslow and is a constant that is unaffected by the conversion. The resolution benefits from a large E value, which in practice means that the selectivity for the wanted enantiomer is high. Since the rate of the indi- vidual enantiomers can be tedious to measure due to the requirement of pure samples of both enantiomers, Eq (1), (2) and (3) have been developed. They allow the E value to be calculated from any two of the following parameters: enantiomeric excess of the starting material (ees), enantiomeric excess of the 21 product (eep) and conversion of the reaction (c).

7 Eq. (1) Eq. (2) / Eq. (3) /

Examples of stereoselective enzymes that have been used within this thesis are the lipases Candida Antarctica Lipase B (CALB) and Lipase PS from Burkholderia Cepacia and the protease Subtilisin Carlsberg.

1.4 Dynamic kinetic resolution and racemization The drawbacks of the 50% maximum yield of the KR approach can be cir- cumvented by the introduction of an in situ racemization catalyst that keeps the substrate racemic throughout the reaction. This is referred to as a dynamic kinetic resolution (DKR) and if the enzyme is highly stereoselective the product can, in theory, be formed in 100% yield with an ee of 100%, (Scheme 4).22

Scheme 4. (R)-selective dynamic kinetic resolution of a sec-alcohol.

If the DKR protocol is applied to a diastereomeric compound the two interconverting species will have different energy, implying that the Keq ≠ 1. The substrates in the enzymatic reaction will have different starting energies and the process is called a kinetic asymmetric transformation (KAT) instead of KR. The inversion of a chiral center in a diastereomeric compound is called epimerization and the combination of epimerization and KAT is termed dynamic asymmetric transformation (DYKAT).23 There are several types of DYKAT reported and in this thesis a variant depicted in Scheme 5 has been employed.

Scheme 5. (R)-selective DYKAT of a diastereomeric substrate.

Depending on the compound that needs to be racemized, there are different methods available. Some examples of racemization are: (i) thermal racemization, (ii) base and acid-catalyzed racemization, (iii) racemization via Schiff-base intermediates, (iv) enzyme-catalyzed racemization, and (v) racemization via reduction-oxidation (redox) reactions.24 In this thesis two homogenous organometallic and one heterogenous nano-structured catalysts, working via hydrogen transfer mechanisms (redox-type racemization) have been used to achieve racemization of sec- alcohols and primary amines. The catalysts abstract a hydride from the chiral alcohol or amine, thereby forming a prochiral ketone or imine intermediate. The alcohol or amine is then reformed by hydride re-addition, and since this occurs in an achiral fashion the stereochemical outcome is random (Scheme 6). Repetition of this process enables racemization of enantiomerically pure sec-alcohols and primary amines.

HX H R R' -[M]

[M] X + R R' + [M] H +H -[M] prochiral H XH [M] R R' X : O, NR'' Scheme 6. General principle of a redox racemization.

There are numerous metal complexes of ruthenium, iridium, palladium, and rhodium developed, capable of performing the hydrogen transfer reaction required to achieve racemization.25 However, not all of them are suitable for chemoenzymatic DKR. To ensure a well-functioning process the following criteria have to be fulfilled:

9 (i) the racemization catalyst must be compatible with the enzyme, (ii) there must be no racemization of the product, (iii) the racemization rate should be at least equal, and preferably faster, compared to the rate of conversion of the fast reacting enantiomer (i.e krac≥kfast). However, if the stereoselectivity of the resolution catalyst is high, it is 23,26 sufficient that krac is much greater than kslow. A few racemization catalysts fulfilling these criteria have been developed for the racemization of sec- alcohols and some of the processes have also found industrial applications (Figure 6).27,28 The first example was published by Williams and co- workers, who achieved racemization with rhodium acetate; however, only a moderate yield of 60% was obtained.29 Shortly after the Bäckvall group reported a DKR with high yield and excellent ee, employing the Shvo catalyst (4) for racemization.30 Further improvement of the DKR was possible after the development of the monomeric catalysts 5-9, which were easier to combine with the enzymes due to their high activity at ambient temperature.31 Catalyst 4 and 5 are used in this thesis and will be discussed in more detail.

Figure 6. Examples of homogeneous, enzyme-compatible racemization catalysts.

1.4.1 Shvo’s catalyst The Shvo catalyst (4) is a dimeric organoruthenium compound used in many organic reactions including oxidations, reductions and racemizations.32 Cata- lyst 4 has successfully been employed in combination with enzymes for the DKR of both alcohol and amine substrates.27a Upon heating in solution the dimer splits heterolytically into the two monomeric ruthenium entities 4a and 4b (Scheme 7). Complex 4b is a 16-electron species that acts as a dehydrogenation catalyst, which transforms a sec-alcohol into a ketone while interconverting itself into 18-electron complex 4a. 4a then acts as a hydro-

10 genation catalyst that in an achiral fashion reduces the ketone back to the alcohol simultaneously reforming 4b. It is generally accepted that the dehydrogenation of alcohols with catalyst 4 operate via an outer-sphere mechanism, which means that the ketone intermediate formed does not stay coordinated to the metal complex. Instead it might dissociate from the coordination sphere if re-addition of hydride is not sufficiently fast.33 This might lead to formation of oxidized byproducts in reactions using 4 as racemization catalyst. Another drawback of the racemization mediated by catalyst 4 is its requirement of elevated temperatures in order to be activated. Consequently, the enzymes combined with 4 must be thermostable.

Ph Ph Ph O O Ph Ph H Ph OH O Ph Ph Ph Ph Ph Ph Ph Ph H Ph Ru Ru Ph Ru Ru OC CO OC H CO CO OC OC CO 4 4a 4b Ph Ph OH Ph Ph O Ru O OC H R R 1 2 OC R1 R2 4a

OH OH Ph Ph O R1 R2 R1 R2 Ph Ph Ru

OC CO

4b Scheme 7. Racemization via hydrogen transfer catalyzed by 4a and 4b.

1.4.2 Racemization catalyst 5 Catalyst 5 is an 18-electron ruthenium complex that has been shown to be able to racemize 1-phenylethanol completely within 10 minutes at room temperature.31 The catalyst has also been successfully applied in many DKR processes of sec-alcohols.34 The proposed racemization mechanism is shown in Scheme 8.35 The active species, 5a, is obtained after nucleophilic displacement of the chloride ligand with a t-butylalkoxide anion (Scheme 8, step i). This activation can easily be observed when performing the reaction,

11 since the reaction mixture immediately changes color from yellow to orange upon addition of t-BuOK. Subsequent alcohol-alkoxide exchange forms the ruthenium sec-alkoxide complex 5b (step ii). The substrate stays coordinated to the metal while a CO dissociation enables a β-hydride elimination to form ketone intermediate 5c (step iii). This is followed by insertion of the ketone into the Ru-hydride bond which forms alkoxide 5d (step iv). The re-addition of the hydrogen can occur at either face of the ketone intermediate, thereby leading to racemization of the alcohol. After a ligand exchange, the rac- alcohol is exchanged by another sec-alcohol and the cycle can start all over again (step v). Until recently, it was debated whether the coordination site, needed in the formation of the ruthenium hydride (5c), proceeded via a CO dissociation or a ring-slippage from η5 to η3. Recent results, based on both theoretical and experimental studies support the CO dissociation.35c-d

Scheme 8. Proposed catalytic cycle for the racemization of sec-alcohols by 5.

12 1.5 General concerns regarding DKR A common issue when designing a chemoenzymatic DKR reaction is to find an optimal rate for both the enzymatic process and the racemization. One of the most straightforward methods to alter the rate and enantioselectivity of enzyme-catalyzed reactions are adjustments of the solvent and its water activity. However, due to requirements of the racemization catalysts, these alterations are not always allowed in a DKR. For example, catalyst 4 and 5 are air-stable and can be stored at room temperature, but their activated catalytic species formed during catalysis are highly sensitive to water and oxygen. The reactions require an inert atmosphere and dry, non-polar solvents, preferably toluene, in order to achieve a well functioning racemization. The- se requirements limit the engineering of the reaction conditions to mainly adjustment of the catalysts loading and alterations of the reaction temperature. For catalyst 5, which is more sensitive than 4, a few precautions have to be made. The substrate requires pre-drying with molecular sieves, the acyl donor requires distillation and storage over MS 4Å. Furthermore Na2CO3, dried under vacuum at 200 °C, needs to be added to the reaction. Na2CO3 is postulated to have two effects; extra drying of the solvent and buffering of any acid formed via acyl donor hydrolysis.36 Catalyst 4 on the other hand where the intermediate ketone may dissociate from the catalyst, sets some restriction on the use of acyl donor. In order for 4a not to react with anything else but the ketone intermediate, no hydride acceptor can be present in the reaction. This excludes the use of acyl donor such as isopropenyl acetate (1), since acetone is formed as a byproduct. In- stead catalyst 4 must be used in combination with one of the acyl donor releasing an alcohol with low nucleophilicity such as p-chlorophenyl acetate (Scheme 3, 2).

1.6 Objectives of this thesis The general efficiency of the DKR using hydrolases and transition metal- complexes has increased constantly over the past decade, and the level of substrate complexity has increased accordingly. The aim of this thesis is to further explore the limits of the DKR protocol and incorporate novel structural motif. The majority of the projects describe the application of the DKR protocol on various substrate classes. Focus is put on obtaining high yields and ees while keeping reaction conditions as mild as possible. Rather than performing reactions on substrates with highly similar substitution patterns we make an effort to subject synthetic interesting substrates to the method. Enzymes play a crucial role in all reactions performed in this thesis; however, their selectivity is not always fully understood. In the last project we

13 study CALB-catalyzed reactions in hydrous and anhydrous environments. The two reactions are postulated to proceed with different levels of enantioselectivity and we try to rationalize why this difference occur.

14 2. Dynamic Kinetic Resolution of Cyclic 1,2 Amino Alcohols

2.1 Introduction Chiral 1,2-amino alcohols are versatile intermediates used in the synthesis of several biologically active substances, unnatural amino acids and chiral auxillaries.37 The sub-class of N-heterocyclic alcohols are found in a wide range of biologically active alkaloids. Some examples are ROCK-inhibitors, targeting several cardiovascular diseases38 and 3-hydroxypyrrolidinols that can be converted into the Geisman-Weiss lactone; a common precursor to various alkaloids in the necine base family.39 Optically active N-heterocycles can also be used as chiral ligands in organometallic complexes or as organocatalysts (Figure 7).37

Figure 7. Examples of N-heterocycles derived from 3-piperidinol and 3- pyrrolidinol.

Various methods have been described for the preparation of N-heterocyclic alcohols in an optically active form,40 however there are only scarce examples of enzyme-catalyzed kinetic resolutions and they generally require high loadings of biocatalyst.41 Moshida and co-workers41b published a kinetic resolution protocol where benzyl-protected 3-pyrrolidinol 11a in a two-step synthesis was transformed into enantiomerically enriched acetate 12a in 83% yield and 91% ee (Scheme 9)

Scheme 9. Moshida’s method for preparing enantiomerically enriched N-substituted 3-acetoxypyrrolidine.

We envisioned that by transforming this KR protocol into a DKR protocol both the yield and the product ee could be increased, making the preparation of 12a more efficient. Also, it would be interesting to expand the scope of the reaction to include both pyrrolidinols and piperidinols with various N- substituents.

2.2 Results and discussion

2.2.1 Optimization of KR and DKR of 11a and 13a The KR of 13a was optimized in toluene at room temperature employing isopropenyl acetate (1) as acyl donor (Table 1). As previously shown,40c non- immobilized lipase PS was very slow in catalyzing the transesterification (Entry 1). When using lipase PS immobilized on ceramic beads or celite, marked as PS-C and PS-IM respectively, the activity increased but only moderately enantioselective reactions were obtained (Entries 2-3).

Table 1. Kinetic resolution of 13a with various enzymes.a

conv. ee ee Entry Enzyme OH OAc Ed (%)b (%)c (%)c 1 PS 0 - - - 2 PS-C 13 13 89 20 3 PS-IM 19 16 91 25 4 IL1-PS 48 88 98 >200 (290) 5 CALB 51 96 89 67 a) Reaction conditions: 13a (0.25 mmol), enzyme (20 mg/mmol), isopropenyl acetate (1), (2 equiv) were stirred in toluene (1 mL) for 20 h. b) Calculated from c=ees/(ees+eep). c) Determined by chiral HPLC. d) Calculated from the eeOH and eeOAc according to Eq. (3). Value given after considering the maximal measurement error, exact number in parenthesis.

16 Interestingly, the same enzyme coated with an ionic liquid, IL1-PS,42 catalyzed the transesterification in a fast and highly selective manner (Table 1, Entry 4). This highlighted the impact the immobilization technique can have on the performance of the lipase. CALB also was found to catalyze the transesterification of 13a at a high rate; however, the selectivity obtained was considerably lower than that of IL1-PS (Table 1, Entry 5). A DKR of 13a, employing Shvo’s catalyst (4) for the racemization and IL1-PS for the resolution, was performed at 50 °C. The reaction did not proceed to full conversion within 20 h and the ee of 14a was only 86% (Table 2, Entry 1). Increasing the temperature to 70 °C allowed the DKR to proceed in full conversion to the acetate with an ee of 91% (Entry 2). The reason for the increase in ee was argued to be due to the increased racemization rate of 4 at elevated temperatures. Further elevation of the temperature was not possible since lipase PS is known to decompose at temperatures above 70 °C.43 In- stead a change of racemization catalyst to monomeric Ru-complex 5 was made, with the aim of obtaining a more efficient racemization. At 70 °C the DKR with complex 5 in combination with acyl donor 1, Na2CO3, and IL1-PS fully converted 13a into 14a in 94% ee (Entry 3). By lowering the temperature to 60 °C the ee could be increased to 96% (Entry 4).

Table 2. DKR of model substrates 11a and 13a.a

Temp. Conv. ee Entry Substrate Ru-cat. (°C) (%)b (%)c 1 4 70 >99 90 2 4 60 >99 91 3 5 70 >99 94 4 5 60 >99 96

5 4 60 >99 4 a) Reaction conditions Shvo’s catalyst (4): sec-alcohol (0.5 mmol), Ru-cat. 4 (2.5 mol%), IL1-PS (40 mg/mmol), p-chlorophenyl acetate (2) (3.5 equiv) stirred in dry toluene (1 mL) 5 under Ar-atmosphere. Reaction conditions Ru(CO)2Cl(η -C5Ph5) (5): sec-alcohol (0.5 mmol), Ru-cat. 5 (5 mol%), t-BuOK (6 mol%), Na2CO3 (0.5 mmol), IL1-PS (25 mg/mmol), isopropenyl acetate (1) (2 equiv) in dry toluene (1 mL). b) Determined by 1H NMR c) Determined by chiral GC or chiral HPLC

In parallel with the optimization of the DKR of 13a, the DKR of substrate 11a was investigated using catalyst 4. However, the ee was as low as 4% and an oxidized byproduct, N-benzylpyrrole, had been formed in approximately

17 10% yield (Entry 5). The low ee was surprising, and slow racemization of 4 alone could not account for this drop in enantioselectivity. In order to elucidate the poor ee, 11a was subjected to a small enzyme screening, which showed that IL1-PS displays low enantioselectivity for substrate 11a (Table 3, Entry 2). PS-IM and PS-C1 on the other hand were highly stereoselective for 11a, once again highlighting the remarkable effect the immobilization method can have on the enantioselectivity of the enzymatic reaction.

Table 3. Comparison of the E values obtained in the KR of 11a and 13a.a E valuesb Entry Substrate PS-C1 PS-IM IL1-PS

1 20 25 290

2 115 >600 11 a) Reaction conditions: sec-alcohol (0.25 mmol), enzyme (20 mg/mmol), isopropenyl acetate (1) (2 equiv) stirred in toluene (1 mL) for 20 h. b) Calculated from the eeOH and eeOAc according to Eq. (3). ee for all reactions are displayed in Table 4

2.2.2 Substrate scope The scope of the enantioselective transesterification reaction was investigated by varying the N-substituents of both 3-pyrrolidinol and 3-piperidinol. Because of the variation observed in enantioselectivity for different lipases in KR of 13a and 11a every substrate had to be screened in detail, in order to find its ideal biocatalyst (Table 4). PS-IM and IL1-PS were both highly enantioselective for isopropyl- and cyclohexyl N-substituted 3-piperidinols 13b and 13c (Entry 5-10). With a cyclohexylmethyl substituent on the nitro- gen (13d), both PS-IM and IL1-PS lost all catalytic activity while CALB displayed high selectivity and high activity (Entries 11-13). All investigated enzymes displayed good enantioselectivty for substrate 13f (Entry 14-16), with CALB being the most active. PS-IM was the most stereoselective enzyme for Cbz-protected 3-pyrrolidineol 11b (Entry 20-21). IL1-PS was best for 11c; however, the E value was only moderate (Entry 22-24). No clear trend regarding which factors that influence the enantioselectivities could be delineated from the results obtained in the enzymatic screening. Instead the results highlighted the importance of investigating each cyclic amino alcohol for its most suitable lipase.

18 Table 4. KR of N-heterocyclic alcohols employing different lipasesa

Conv. ee ee Entry Substrate Enzyme OH OAc Ed (%)b (%)c (%)c 1 PS 0 - - - 2 PS-IM 19 16 91 25 3 IL1-PS 48 88 98 >200 (290) 4 CALB 51 96 89 67 5 PS-IM 18 20 99.7 >400 (810) 6 IL1-PS 11 9 99.4 >200 (363) 7 CALB 41 64 96 95 8 PS-IM 26 36 99.0 >200 (283) >600 9 IL1-PS 48 93 99.4 (1143) 10 CALB 50 94 93 98 11 PS-IM 5 nd - -

12 IL1-PS 5 nd - - 13 CALB 42 nd 98 >200 14 PS-IM 9 12 99 200 15 IL1-PS 34 50 99 >200 (327) 16 CALB 47 90 99 >300 (617) >600 17g PS-IM 20.5 nd 99.8 (1285) 18 IL1-PS 65 nd 52 11 19 CALB 100 - - - 20f PS-IM 45 80 >98 >200 (244)

21f IL1-PS 55 99.8 82 67

22 PS-IM 20 23 94 40

23 IL1-PS 51 92 90 62 24 CALB 73 46 - 10 a) Reaction conditions: sec-alcohol (0.25 mmol), enzyme (20 mg/mmol), isopropenyl acetate (1) (2 equiv) stirred in dry toluene (1 mL) for 20 h. b) Calculated from the formula c=ees/(ees+eep). c) Determined by chiral HPLC or chiral GC. nd = not determined. d) Calculated from the eeOH and eeOAc according to Eq. (3). Value given after considering the maximal measurement error, exact number in parenthesis.

19 2.2.3 Dynamic Kinetic Resolution With the knowledge of which enzyme to use for every substrate the KRs were expanded into DKRs. DKR of 13b was attempted at 50 °C using 40 mg/mmol of PS-IM. Unfortunately, full conversion was not reached within 24 h and the ee was slightly lower than expected (Table 5, Entry 2). In order to increase the rate of the transesterification the temperature was increased to 60 °C in combination with larger amounts of enzyme (80 mg/mmol) and acyl donor (4 equiv). This provided 14b in 76% yield and in 95% ee (Entry 3). By performing the reaction at 50 °C instead of 60 °C the enzyme selectivity was enhanced and an ee of 96% could be obtained (Entry 4). Increased amounts of acyl donor and enzyme were required to ensure an efficient DKR of all substrates bearing an aliphatic N-substituent (13b-13e). N-cyclohexyl-3-acetoxy-piperidine, 14c, was isolated in an 89% yield and an excellent ee of 99% and its unsaturated analog 14d was converted in 83% yield with an ee of 99%44 (Entry 5-7). The resulting conversion and ee of 14e were lower than expected under the optimized conditions and further modification of the reaction conditions failed to improve the enantioselectivity (Entry 8). 1-Tosylpiperidin-3-yl acetate 14f was obtained in 88% yield and 96% ee (Entry 9). This result shows that both the enzyme and racemization catalyst are compatible with compounds having several heteroatoms. N-Benzyl and Cbz protected 3-pyrrolidinol 11a and 11b were efficiently converted to the enantioenriched acetates 12a and 12b, respectively, in yields above 80% and more than 95% ee (Entry 10-11). The DKR of 1-phenyl-3-pyrrolidinol (11c) was first conducted with IL1- PS, since this enzyme displayed the highest stereoselectivity in the KR study. The result was discouraging though, with a product ee of 28% (Entry 12). However, PS-IM, albeit being less selective and slower, was able to deliver the desired acetate in 82% yield and 86% ee (Entry 13).

2.3 Conclusion The efficiency of the enzyme-catalyzed preparation of 12a has been increased by employing a DKR protocol instead of a KR protocol. We found that less enzyme could be used by employing immobilized enzyme prepara- tions. Interestingly the method of immobilization had an impact on the enantioselectivity of the enzyme. The DKR protocol was also successfully applied to various 3-pyrrolidinols and 3-piperidinols equipped with both electron-withdrawing and electron-donating N-substituents. Compounds 12a, 12b, 14a and 14f are of special interest since these potentially can be deprotected, forming unsubstituted cyclic 1,2-amino alcohols.

20 Table 5. Dynamic kinetic resolution of substrates 13a-f and 11a-c.a

Enzyme 1 Yield ee Entry Product (x mg/mmol) (equiv) (%)b (%)c

1 IL1-PS (25) 2 91 96

2d PS-IM (40) 2 (84) 92 3 PS-IM (80) 4 76 95 4d PS-IM (80) 4 86 96

5d PS-IM (80) 6 89 99

6d PS-IM (80) 4 (70) nd

7d PS-IM (80) 6 (83) 99e

8 CALB (40) 4 73 90

9f CALB (25) 2 88 96

10 PS-IM (25) 2 80 95

11 PS-IM (25) 2 87 95

12 IL1-PS (40) 2 (>99) 28

13 PS-IM (25) 4 82 86 a) Typical conditions: sec-alcohol (0.5 mmol), Ru-cat. 5 (5 mol%), t-BuOK (6 mol%), Na2CO3 (0.5 mmol), lipase (25-80 mg/mmol), isopropenyl acetate (1) (2-6 equiv) were stirred at 60 °C for 20 h. b) Isolated yield. Value in parenthesis shows conversion based on 1H NMR. c) Determined by chiral GC or chiral HPLC. d) Reaction performed at 50 °C. e) ee determined after hydrogenation of a 1:1 mixture of diastereomers into enantioenriched 13c. f) 6 mol% of Ru-cat. 5 used.

21 3. DKR and DYKAT of Functionalized Cyclic Allylic Alcohols. Application to Glycal Derivatives

3.1 Introduction Cyclic allylic alcohols are synthetically useful intermediates that can be further functionalized through transformations such as epoxidations, SN2’ reactions, and Claisen rearrangements (some selected examples are shown in Scheme 10). If the alcohol is enantiomerically enriched the chirality could in theory be transferred to the product and also used to induce chirality into a new chiral center.4547

Scheme 10. Examples of possible functionalization of cyclic allylic alcohols

Optically active allylic alcohols have traditionally been prepared by asymmetric reduction of α,β-unsaturated ketones, enantioselective addition of alkenyl metal reagents to aldehydes or the enantioselective addition of alkyl metal reagents to 2-enals.4850 There are also a few examples of chemoenzymatic DKR of racemic allylic alcohols, affording enantioenriched allylic acetates.51 Racemization has mainly been achieved by either oxo transition-metal reagents, capable of forming racemic allyl cations, or

22 organoruthenium catalysts working via redox mechanisms.52 Common problems observed in the redox processes are the competing oxidation and isomerization reactions, which respectively, forms a α,β-unsaturated ketone or a saturated ketone as byproducts (Scheme 11). The formation of oxidized byproducts is associated with the conjugated enone intermediate formed in the oxidation step of the racemization. The conjugation makes the intermediate more stabile and this will increase its barrier for reduction. The isomerization has been found to be more severe if the allylic moiety is on a terminal position and if the enone intermediate can adopt an s-cis conformation, since this can bring the olefin in a favorable geometry for a 1,4-hydride addition.53 The aim of this project was to extend the scope of the lipase- and ruthenium-catalyzed DKR to also include C3 functionalized cyclic allylic alcohols. An initial concern was that both catalyst 4 and 5 have been shown to be efficient isomerization catalysts for allylic alcohols; however, since the racemization usually is faster than the isomerization we believe that the amount of isomerized byproducts could be constrained.54 Moreover, monomeric catalyst 5 does not perform the 1,4-hydride addition on medium-sized cyclic substrates due to the distance between the metal hydride and position 4 in the intermediate s-trans enone.55

Scheme 11. Side reactions leading to byproducts in the racemization of allylic alcohols.

3.2 Results

3.2.1 Kinetic resolution Before the DKR was studied the enzymatic stereoselectivity for cyclohex-2- enols with 3-substituents of different sizes was measured. The reactions were carried out in toluene at room temperature, employing isopropenyl acetate, 1, as acyl donor. Alcohol 15a was well resolved by both CALB and PS-IM (E

23 value 160 and 170, respectively) at a moderate rate, with the former being the faster. AK Amano 20, PS-C1 and IL1-PS all delivered 16a at a low rate, and therefore these enzymes were not further evaluated. With smaller substituents at C3 the enantioselectivity displayed by the enzymes decreased; for instance CALB had an E value of 63 for 3- methylcyclohex-2-enol (15b) and 2.2 for cyclohex-2-enol (15c). These results are nonetheless interesting since they show that even a minor size difference such as that between a saturated and an unsaturated bond makes it possible for the enzyme to display enantioselectivity (Table 6).

Table 6. KR of 3-substituted cyclohex-2-enols.a

Time ee ee Entry Substrate Enzyme conv.b OH OAc Ed (h) (%)c (%)c 1 CALB 4 46 83 97 160 2 PS-C1 4 8 nd nd nd 3 IL1-PS 4 1 nd nd nd 4 PS-IM 4 23 34 98 170 5 AK Amano 20 4 4 nd nd nd 6 CALB 1 22 27 96 63

7 PS-IM 1 7 5 82 11

8 CALB 1 40 20 30 2.2d

9 PS-IM 1 65 31 17 1.8d a) Reaction conditions: 15 (0.2 mmol), isopropenyl acetate (1) (2 equiv) and enzyme (25 mg/mmol) stirred in toluene (0.5 mL) at r.t. nd = not determinded, due to insufficient conversion. b) Determined by 1H NMR. c) Determined by chiral GC or HPLC. d) Calculated by eeOH and eeOAc according to Eq. (3). d) Reaction is (R)-selective.

3.2.2 Optimization of the DKR Compound 15a was studied in a DKR employing catalyst 5 for racemization and CALB for the resolution. After 24 h at room temperature the reaction had reached 86% conversion, with 16a formed in 47% in a moderate ee of 86% (Table 7, Entry 1). As predicted, no isomerization product was detected, instead the remaining material, were identified as 10% oxidized product 17a and 29% ether compound 18a (in a diastereomeric mixture). With the

24 aim of reducing the byproduct formation the reaction was repeated at 40 °C; however, the result was unsatisfactory by displaying even lower conversion and ee of (R)-16a and a considerable amount of 18a (Entry 2). The origin of 18a was not clear, but previous studies on electron-rich 3-aryl-cyclohex-2- enols have shown that these compounds can undergo spontaneous ether formation at room temperature.56 However, compound 15a could be stored in toluene at room temperature for several weeks without any detectable formation of 18a. Clearly, one or several of the components in the DKR in- duced this undesired background reaction. In order to elucidate which com- ponent that facilitates the formation a racemization test was conducted with only (S)-15a and the activated ruthenium catalyst 5. Within 1 h all of the alcohol had been converted into 18a, showing that 5 catalyzed this formation and that a change of racemization catalyst was necessary.

Table 7. Optimization of the dynamic kinetic resolution of 3-phenylcyclohex-2-enol (15a) with lipase Ru-cat. 4 or 5.a

OH OAc O Ru-cat. 4 or 5 (5 mol%) O Ph acyl donor (2 equiv) toluene Ph Ph Ph Ph 15a 16a 17a 18a Temp ee b 15ac 16ac 17ac 18ac Entry Cat. Enzyme OAc (°C) (%) (%) (%) (%) (%) 1 5 CALB (25 mg/mmol) r.t. 86 14 47 10 29 2 5 CALB (25 mg/mmol) 40 60 22 43 10 25 3 4 CALB (25 mg/mmol) 60 80 - 90 10 - 4 4 CALB (25 mg/mmol) 70 80 - 88 12 - 5 4 CALB (25 mg/mmol) 80 84 - 86 14 - 6 4 CALB (2 mg/mmol) 60 90 - 74 26 - 7 4 PS-IM (50mg/mmol) 60 97 - 91 9 - 8 4 PS-IM (50mg/mmol) 60 96 - 92 8 - 9d 4 PS-IM (50mg/mmol) 60 95 17 70 13 - 10e 4 PS-IM (50mg/mmol) 60 94 12 62 26 - 11f 4 PS-IM (50mg/mmol) 60 91 - 86 14 - 5 a) Conditions using Ru(CO)2Cl(η -C5Ph5) (5): 15a (0.25 mmol), Ru-cat. 5 (5 mol%), t-BuOK (5 mol%), Na2CO3 (0.25 mmol), lipase, isopropenyl acetate (1) (2 equiv) were stirred in dry toluene (1 mL) for 24 h; Conditions using Shvo’s catalyst (4): 15a (0.25 mmol), Ru-cat. 4 (5 mol%), lipase, p-chlorophenyl acetate (2) (2 equiv) were stirred in dry toluene (1 mL) for 24 h. b) Determined by chiral GC. c) Determined after 24 h by NMR analysis. d) Performed in cyclohexane. e) Performed in acetonitrile. f) Performed in diisopropyl ether.

Thus, we turned our attention to catalyst 4 and carried out a DKR keeping the CALB loading at 25 mg/mmol and increasing the temperature to 60 °C.

25 The reaction reached full conversion within 24 h with a moderate product ee of 80%. Gratifyingly, no ether or isomerization byproducts had been formed and the amount of ketone 17a was the same as observed in the DKR with 5 (Table 7, Entry 3). Monitoring the reaction by GC revealed that the substrate ee was not racemic during the reaction, suggesting that the rate of the racemization was insufficient and did not keep up with the transesterification rate. Attempts to increase the rate of the racemization by performing the DKR at 70 or 80 °C did not improve the ee considerably (Entry 4-5). Lower- ing the enzyme loading did raise the ee to 90%, but the slower reaction led to increased amounts of ketone (Entry 6). By employing the more selective and slower resolution catalyst PS-IM in a DKR at 60 °C the conversion of 16a could be increased to 91% in a high ee of 97% (Entry 7). Addition of a hydride donor or performing the reaction in other solvent than toluene did not improve the conversion or the product ee (Entry 8-11).

3.2.3 Scope of the reaction Allylic alcohols 15b and 15d-15h were synthesized according to literature procedures.57 A DKR of 3-methylcyclohex-2-enol (15b) employing CALB for the resolution delivered (R)-15b in full conversion, but only in a moderate ee of 60%. As indicated by the kinetic resolution, the lipases displayed a low stereoselectivity for substrates with smaller substituent at C3, explaining the result. DKR with larger electron-deficient substituents in the vicinal position, represented by a p-chlorophenyl, proceeded smoothly converting 15d to 16d in 91% yield and with an ee of 96% (Scheme 13).With a p- methoxyphenyl at C3 (15e), a large amount of ether was formed and the ee of the acetate obtained was 0%. Both the formation of 18e and the absence of optical activity might be explained by the formation of a stabilized carbo- cation (Scheme 12).

Scheme 12. Plausible explanation of the formation of racemic 16e product and the formation of ether 18e in the DKR of 15e.

Scheme 13. Substrate scope for the dynamic kinetic resolution of cyclic allylic alcohols. Method A employs the Shvo catalyst (4) and Method B employs 5 1 Ru(CO)2Cl(η -C5Ph5) (5). The conversion was measured by H NMR with remaining material consisting of oxidization product 17 unless otherwise noted. Isolated yields in parentheses. n.i. = not isolated. a) CALB (5 mg/mmol) used instead of PS- IM and 2,6-dimethylpentan-4-ol (1 equiv) added. b) 31% ketone, 28% ether 18e.

Transformation of 15f into sulfonyl-substituted cyclohex-2-enyl acetate 16f was achieved in full conversion with an ee of 94%. Compound (R)-16f is highly interesting due to the electrophilic site that the sulfonyl creates in position 2 of the cyclic enol. This structural feature has been used extensively in the preparation of a wide range of functionalized molecules.58 If the DKR of 3-iodocyclohex-2-enol (15g) was stopped at 82% conversion, 11% of ketone and 71% of acetate 16g in 96% ee had been formed. Prolonged reaction times led to lowered ee and increased amounts of oxi-

27 dized byproducts. Further optimizations using racemization catalyst 4 failed to improve these results; however, by changing to catalyst 5 an efficient DKR was obtained reaching 16g in full conversion, with no detection of ketone or ether byproducts, and ee over 99%. 3-Undecylcyclohex-2-enol (15h), a precursor for the total synthesis of the anitfungal Tanikolide,59 was converted to 16h in 87% yield and with an ee of 91%. Two cyclopent-2-enol compounds were investigated. 3- (phenylsulfonyl)cyclopent-2-enol (15i) provided 16i in a moderate ee of 88% and with 40% of enone side product. Given that the E value of PS-IM for 15i was 150, we came to the conclusion that the racemization was much slower than the rate of the enzymatic reaction for this electron-deficient substrate. When 15j was subjected to the DKR conditions, enone formation was completely avoided, but 16j was only obtained in an ee of 26%. The low ee was explained by a E value of only 30 in combination with a slow racemization.

3.2.4 Synthetic application of 16g Because of the high versatility of compounds bearing a vinyl iodine, and since 16g was isolated in excellent ee we were inspired to perform further reactions on this substrate. Inspired by a literature procedure we conducted a Stille cross-coupling reaction installing a vinyl group at C3.60 Gratifyingly, the ee of 16g was fully retained in product 19. Also, a lithium-halogen exchange was possible, which afforded a vinyl lithium species that could be trapped with various electrophiles.61 We showed one example where the lithiated species 20 was quenched by water yielding enantiopure, cyclohexen-3-ol, 15c in 77% yield. This result was rewarding, since 15c initially could not be prepared in high ee using lipases; however, the detour through 16g enabled its preparation in perfect ee.

Scheme 14. Alteration of substrate 16g by Stille cross-coupling yielding 19 and by reduction of the iodine yielding compound 15c. In both reactions the stereochemistry is fully retained.

28 3.3 DYKAT on carbohydrate derivatives

3.3.1 Background Carbohydrates are most often acquired as pure stereoisomers from natural sources and the orientation of the functional groups influnce the chemical and biochemical properties of the carbohydrate. By inverting the chiral centers of readily available carbohydrates it is possible to reach rare or even non-natural occurring variants of carbohydrates.62 Inversion of non-anomeric carbons with hydroxyl substituents regularly follows two-step procedures such as triflation followed by nucleophilic displacement or oxidation and subsequent stereoselective reduction.63,64 An alternative procedure would be to perform a DYKAT reaction where a chiral center of a carbohydrate is in situ epimerized and an enzyme stereoselectively acylates the epimer of the starting material (Scheme 15). This would constitute a highly attractive one- step method for performing an inversion of hydroxyl substituted chiral center and reaching rare carbohydrates. Enzymes are known to successfully distinguish between sec-alcohols on carbohydrates in both transesterification and hydrolysis reactions. The stereopreferences are the same as for any sec-alcohol with lipases preferen- tially acylating the (R)-epimer, whereas proteases are selective for the (S)- epimer (vide supra).65

Scheme 15. General principle of DYKAT of carbohydrates leading to hydroxyl inversion.

The metal-catalyzed epimerization of carbohydrates is less studied. Since the two epimers have different free energy the epimerization does not necessari- ly equilibrate into a 50:50 ratio between the two species. Metal-catalyzed epimerization instead enables an experimental way of addressing the position of equilibrium between the carbohydrate epimers and thereby determine

29 their relative stabilities. Catalyst 5 has been utilized in such a study of partially protected glucose, mannose and allose derivatives. The protocol employed 5 mol% of activated 5 and required a reaction temperature of 120 °C for 24-120 h to reach equilibrium. This high temperature disqualified this epimerization to be combined with an enzyme-catalyzed reaction.66 Catalyst 5 is usually excellent for the epimerization of sec-alcohols at room temperature; however the electron deficient groups present on carbohydrates renders the sec-alcohols to have a high energy barrier for the oxidation step, retarding the epimerization. Furthermore, the steric bulk around the sec-alcohol was large, inhibiting the formation of the ruthenium-alkoxide intermediate, further decreasing the rate of epimerization. By modifying the carbohydrate substrates to their protected glycal analogues, the number of bulky electron withdrawing groups would be reduced. In this study we investigated the epimerization and DYKAT of stereomerically pure carbohydrate derivatives 21, 22 and 23 having a free hydroxyl group at the chiral center of C3 and the remaining hydroxyl groups protected as benzylidene acetals or benzyl ethers (Figure 8).

Figure 8. Substrates investigated in the DYKAT protocol leading to chiral center inversion.

3.3.2 Epimerization and DYKAT of compound 21 and 22 Initially epimerization of 21 in toluene was tested employing complex 4, 5 and 10 (Figure 6). All have previously shown to racemize allylic sec- alcohols; however, in the epimerization of 21 only 5 showed activity at room temperature. An equilibrium of 2:1 of 21 and protected allal 22 was established after less than 30 minutes.51,67 1 By monitoring the epimerization by H NMR experiments in toluene-d8 at 25 °C using 10 mol% of 5 it was observed that the equilibrium between 21 and 22 was reached after less than 5 min. The quantities of the different species were observed by 1H NMR and apart from 21 and 22, the ruthenium- alkoxide intermediates 21* and 22* were detected (Figure 9).68 From the integrals of these species it was possible to determine the equilibrium constants K2 and K1*K3 to 0.56 and 0.90, respectively.

30 21 22

22* 21*

4.8 4.7 4.6 4.5 4.4 1H /ppm

Figure 9. Above. Equilibrium between compounds 21 and 22 using 10 mol% of 5, H2 shown for clarity; Below. Part of the 1H NMR spectrum at 25 °C of the equilibrium mixture, showing the H2 resonances. Peaks arising from the ruthenium-alkoxide complexes are marked with asterisks ([Ru] = catalyst 5).

Moreover, the rapid establishment of the equilibrium between 21 and 22 enabled the epimerization to be studied using a 1D 1H EXSY NMR experiment.69 Since this experiment requires a very rapid epimerization the temperature had to be increased to 90 °C. The 1D 1H EXSY NMR experiment was performed by selective excitation of H2 on epimer 21 or 22, followed by a mixing time of 3 s (Figure 10). From the results it was possible to perform –1 a full exchange matrix analysis giving the exchange rates k+ = 0.055 s and –1 k– = 0.075 s . The ratio between compounds 21 and 22 equilibrated at 4:3 and the catalyst-substrate intermediates was also visible giving the equilibrium constants K1 · K3 = 0.71 and K2 = 1.10. After a couple of minutes of epimerization at 90 °C the 1H NMR spectrum started to show a build-up of the two ketones 24 and its suggested saturated analogue 25 (Figure 10); both irreversibly formed through oxidation and redox isomerization processes previously described (Scheme 11).53b

31 a

b Ph O Ph O O O O O

22 21 O O c 24 25

4.8 4.7 4.6 4.5 1H /ppm Figure 10. Left. Equilibrium between 21 and 22, facilitated by complex 5 in toluene- 1 d8 solution at 90 °C and detected by: a) 1D H EXSY NMR with selective excitation of the H2 resonance at 4.52 ppm in 21, b) 1D 1H EXSY NMR with selective excitation of the H2 resonance at 4.74 ppm in 22 and c) the corresponding 1D 1H NMR spectrum; Right. Byproducts 24 and 25 observed while studying the epimerization of 21 and 22 by 1H NMR at 90 °C

With an effective epimerization at hand the DYKAT was investigated. Since the stereochemistry at C3 of glucal 21 is of (R)-configuration, an enzyme capable of acetylating its (S)-epimer is needed if an inversion to the allal epimer is desired (Figure 11).

Figure 11. Structural analysis of epimers 21 and 22. Stereochemistry at carbon three marked.

As stated in section 1.2, proteases of the families Subtilisin and Chemotrypsin generally show (S)-selectivity and Subtilisin Carlsberg is a protease that previously has been used in (S)-selective DKR protocols. Un- like CALB and Lipase PS, Subtilisin Carlsberg is not active in toluene and requires an acyl donor with at least 4 carbons in the acyl chain to perform transesterification. Hence, the (S)-selective DKRs has previously been performed in THF with butyrate or valerate esters e.g. 2,2,2-trifluoroethyl butyrate 3 (Scheme 2).70 Moreover, Subtilisin Carlsberg is commercially available as a lyophilized crude enzyme mixture and needs to be treated with surfactants such as octyl β-D-glucopyranoside and Brij56 in order to be catalytically active in organic solvents (Figure 12).

32 HO OH

HO O O HO C16H33 ~10 HO OC8H17 Octyl ß-D-glucopyranoside Brij56 Figure 12. Surfactants used in the treatment of Subtilisin Carlsberg’s.

The surfactant treatment is done by mixing Subtilisin Carlsberg with the surfactants in a phosphate buffer, followed by lyophilization.71 The surfactants form a reversed micelle around the enzymes with the concomitant solu- bilization of small amounts of water, thereby providing a stable aqueous microenvironment. The water in the enzyme preparation in combination with THF, being a more polar solvent than toluene, complicates the combination with the water sensitive catalyst 5. In an initial DYKAT experiment with surfactant-treated Subtilisin Carlsberg (STS) in THF, using 10 mol% of complex 5, 3 as acyl donor and Na2CO3 as drying agent an immediate color change from red to brown was observed. The color change indicated a decomposition of 5. Ana- lyzing the reaction by 1H NMR after 18 h showed 33% of allal ester 26, with the rest of the reaction mixture composed of 21 (Table 8, Entry 1). This result shows that the epimerization catalyst swiftly sets the equilibrium to 2:1 before it decomposes. In the coming hours the enzyme is acylating all the available epimer 22 into 26. Noteworthy is the high diastereoselectivity displayed by the enzyme since no acylation of 21 was detected. Most probably 5 decomposed due to the relatively high water content of the reaction, and to reduce the water content several measures were taken. The THF was distillation over Na and benzophenone, followed by storage over activated molecular sieves (4Å) for 3 h giving it a water content of 10 ppm. The STS and 21 were stored in vacuo for three days with activated molecular sieves (4Å) and the acyl donor was distilled twice. Combining these dried reagents resulted in a DYKAT that after 24 h had proceeded to 26 in 55% conversion, with the remaining reaction mixture consisting of only 21, again showing the high diastereoselectivity of the enzyme (Entry 2).

33 Table 8. DYKAT of glucal derivative 21.

5 Time 26a 27 Entry (mol%) (h) (%) (%) 1b 10 18 33 - 2b 5 24 55 (44) - 3c 7.5 40 78 (71) 7 a) Isolated yield in parenthesis. b) Full amount of 5 present at the beginning of the reaction. c) 2.5 mol% 5 activated by 2.5 mol% of t-BuOK present at the start of the reaction. Two extra batches of 2.5 mol% activated 5 in dry THF (0.1 mL) added after 18h and 36h.

The absence of allal epimer 22 and an observed color change of the reaction from red to dark brown after 4 h again suggested that the catalyst had decomposed. In order to verify the suggested decomposition of complex 5 and simultaneously ensure that Subtilisin Carlsberg was still active (S)-1- phenylethanol was added to the reaction. After 2 h the added sec-alcohol had been converted to its corresponding butyrate in 16% conversion; however, no racemization to give (R)-1-phenylethanol was observed. By adding extra portions of epimerization catalyst after 18 and 36 h the conversion could be increased. At every addition the catalyst survived long enough to reset the 2:1 equilibrium between 21 and 22 making it possible to acquire 26 in 71% isolated yield (Entry 3).

Scheme 16. DYKAT of allal derivative 22.

The reverse epimerization reaction, preparing glucal ester 28 from allal 22 was achieved using CALB as transesterification enzyme. The reaction could be performed in toluene and 83% of 28 was isolated after 18 h reaction time (Scheme 16). No acylation of 22 was observed indicating that also CALB is highly selective in this reaction.

34 3.3.3 Epimerization and DYKAT of compound 23 Both unsaturated carbohydrates 21 and 22 can be prepared from their naturally occurring saturated analogues, making the interconversion between glucal 21 and allal 22 a reaction of mainly academic interest.72 A transformation that would be more useful would be the inversion of the hydroxyl chiral centre of L-Rhamnal derivative, 23, yielding axial compound 29. This epimer can be transformed into L-digitoxose which is a structural element found in various biological active substances such as Jadomycin B73 and Kijanimicin74 (Scheme 17). The synthesis of L-digitoxose usually requires several synthetic steps and setting the stereochemistry of C3 have often shown to be challenging.75

Scheme 17. DYKAT epimerization of 23 would give 29; a precursor for L- digitoxose, which is found in biological active substances.

We started by studying the epimerization of 23 and 30 which was shown to be much slower than that of 21 and 22; requiring four days to reach equilibrium (Figure 13).

100

40 population /% 23

30 20 31 + 32

0 01234 t /days Figure 13. Plot of the epimerization equilibrium between 23 (red squares) and 30 (blue circles) after treating 23 with t-BuOK-activated 5 (10 mol%) in toluene-d8 solution at 25 °C. Oxidation byproducts 31 and 32 (green triangles) are formed in small amounts. Solid lines are drawn through the data points as an aid to the eye. 1 Populations are obtained by integration of H NMR spectra.

35 We hypothesized that the slow epimerization may be ascribed to the steric bulk created by the freely rotating protecting group adjacent to the hydroxyl group. This can potentially interfere in the interaction between Ru-cat. 5 and the hydroxyl group. Supported by the absence of ruthenium-substrate complex in 1H NMR, we concluded that the Ru-alkoxide complexes of 23 and 30 were disfavored compared to the free Ru-OtBu complex. As for allal 22, L-Rhamnal 23 has a sec-alcohol at the C3 chiral center with an (S)-configuration. In order to achieve the inversion of stereochemistry by transesterification of the (R)-epimer, CALB was chosen as enzyme. The reaction was performed in toluene with 5 mol% 5 yielding 81% of 29 with the remaining reaction mixture being constituted of 8% and 11% of 31 and 32, respectively (Scheme 18). Because of the slow epimerization the reaction required 6 days to reach completion.

Scheme 18. DYKAT of substrate 23 with isolated yield of 29 in parenthesis.

3.4 Conclusion Efficient dynamic transformations of various cyclic allylic alcohols have been developed. Optimization of the DKR conditions for C3-functionalized cyclohex-2-enols and cyclopent-2-enols enables the transformation into their corresponding allylic esters in high ee and yields. Byproducts are, controver- sial to that predicted, decreased by employing catalyst 4 instead of 5. How- ever, electron deficient substrate 15g can be combined with complex 5 giving 16g in high yield and perfect enantioselectivity. By performing a cross- coupling and a lithiation reaction the versatility of 16g is demonstrated. The results show the versatility and importance of this substrate. The DKR protocol was extended to also include the first DYKAT of glycal substrates. Partially protected D-glycals 21 and 22, and L-glucal 23 were treated with complex 5 and a hydrolase, acylating the corresponding epimers of each substrate. The outcome of the reaction was an inversion of a hydroxyl substituted chiral center, yielding an epimer of the starting glycal. The epimerization was studied by 1H NMR experiments making it possible to determine equilibrium constants and exchange rates for the unsaturated carbohydrate epimers.

36 4. Dynamic Kinetic Resolution of Primary Amines, using a Recyclable Nanopalladium Catalyst for Racemization

4.1 Introduction Optically active amines and amides are versatile synthetic intermediates that are frequently encountered in a variety of biological active molecules. There exist several techniques for preparing enantioenriched amines, such as optical resolution of racemic amines by crystallization, reductive amination or transamination of prochiral ketones,76,77 hydrogenation or alkylation of prochiral imines and enamines,78 and aminohydroxylation of alkenes.79 Also hydrolase-catalyzed KR, functioning under mild reaction conditions and giving high enantioselectivities, has been used in several studies.15 However, transforming a KR into a DKR of amines has proved to be more challenging than for sec-alcohols. This is due to the increased harshness of the reaction conditions required to achieve the in situ racemization of the amine. Hence, most metal catalysts capable of racemizing amines are more or less inactive at temperatures below 100 °C making the racemization process more difficult to combine with most enzymes.80 In recent years a few DKR systems employing metal-catalyzed racemization have emerged, which have been able to deliver chiral amides in high ees and good yields (Figure 14). The first example of a DKR using transition metal-catalyzed racemization together with a lipase was reported by Reetz and co-workers in 1996. Palladium on carbon was used for the racemization and CALB for the enzymatic resolution of 1-phenylethylamine, 33a, yielding 64% of enantiopure 34a after 8 days of reaction.81 The next advance in the field was when our group reported on a DKR using an electron rich version of Shvo’s catalyst, 4’, for the racemization. The system could be applied to both benzylic and aliphatic amines; however, a temperature of 90 °C was required.36 In parallel to our study, Jacobs and co workers published a catalyst consisting of Pd deposited on basic support. Their Pd nanoparticles were within an average size range of 5-10 nm and the DKR of 33a and other benzylic amines could be performed at 70 °C; however, dilute conditions were required.82

37 Following this study, other catalytic systems using Pd nanoparticles have been developed;83 the most efficient being a system by Kim and Park employing Pd nanoparticles entrapped in an aluminum oxide matrix.84 The high activity of their racemization catalyst was ascribed to the small size of the nanoparticles (average diameter 1.73 nm). This system transformed 33a into enantiopure 34a in 6h at 70 °C. The reaction also worked at 50 °C; however, the ee dropped to 98% and the reaction requires 48 h to reach completion.

Figure 14. Previous advances in the metal- and lipase-catalyzed DKR of racemic amines. The systems employ different racemization catalysts and require dilute conditions and elevated temperatures.

In several of the reported reactions a common problem has been the lowered yield due to a background reaction between the intermediate imine formed in the racemization and a substrate amine. This process forms a dimer that subsequently can be reduced and split to ethylbenzene and an amine. The imine intermediate is also sensitive to moisture and might decompose to the corresponding ketone (Scheme 19). Hence, most DKR reactions have required dilute conditions leading to long reaction times. A further disadvantage of the reported DKRs is that the elevated reaction temperatures have implied that CALB virtually has been the only enzyme used.

Scheme 19. Metal-catalyzed racemization of benzylic primary amines and possible side reactions forming ketones and/or deaminated products.

In order to further improve the DKR protocols of amines, our group developed a heterogeneous catalyst consisting of Pd nanoparticles deposited on amino-functionalized siliceous mesocellular foam (Pd0-AmP-MCF). This MCF support has an amorphous three dimensional network of pores with large internal surface, which protects the palladium nanoparticles from leaching as a result of mechanical grinding. The Pd is added to the material as Li2PdCl4 and then reduced by NaBH4 to form nanoparticles with an average diameter of 2.5 nm.85 The catalyst was tested in the racemization of amines and delightfully the Pd0-AmP-MCF was able to racemize 33a at 50 °C and also to some extent at 40 °C. Later on, the catalyst has also been successfully employed in several other organic transformations, such as oxidation of alcohols, reduction of olefins and nitro compounds, and Suzuki cross-coupling reactions.86 The MCF material could also be used to co-immobilize CALB and Pd nanoparticles in the same cavity, and the resulting hybrid catalyst was used in the DKR of 33a.87 Even though the rate of the reaction was enhanced within the cavity, optimization of the reaction conditions was difficult because of the fixed ratios between the two catalytic species. Because of this and due to the fact the pure Pd0-AmP-MCF catalyst never had been optimized for the DKR of amines we were interested in studying the DKR with this racemization catalyst. The aim was to explore the scope of the reaction in terms of temperature and substrates, incorporate enzymes that are less thermostable than CALB and analyze the recyclability of the system.

39 4.2 Result and discussion

4.2.1 Optimization of DKR From previous studies it is known that CALB shows excellent stereoselectivity for amine 33a, even at elevated temperatures.36 In general, two types of acyl donors have been used, alkyl acetates or alkyl methoxyacetates. The ether oxygen at the carbonyl part of the latter acyl donor is believed to create an extra hydrogen bond in the active site, thereby increasing the nucleophilicity of the substrate amine. This makes the enzyme-catalyzed reaction with alkyl methoxyacetates about 200 times faster than that of their corresponding alkyl butyrate.88 An initial DKR in toluene 0 under H2 gas atmosphere at 70 °C using 2.5 mol% of Pd -AmP-MCF, 2 equiv of ethyl methoxyacetate, 35 (Table 9), as acyl donor and Na2CO3 as drying agent converted 33a into 34a in excellent ee, however only in 63% conversion (Table 9, Entry 1). By concentrating the reaction from 0.15 M to 0.40 M the reaction went to completion within 16 h (Entry 2). A change from Na2CO3 to molecular sieves further decreased the time required to reach full conversion and also made it possible to decrease the palladium loading to 1.25 mol% (Entry 3-4). This low Pd and CALB loading, high substrate concentration and short reaction time makes the reaction more efficient than any of its predecessors. It is also noteworthy that none of the byproducts obtained in several of the previous reported DKR protocols were detected in any of the DKR involving substrate 33a.

Table 9. Optimization of the DKR of 33a at 70 °C.a

Time Toluene Conv. ee Entry Pd-loading Additive (h) (mL) (%)b (%)b

1 2.5 mol% Na2CO3 (1 equiv) 24 4 63 99

2 2.5 mol% Na2CO3 (1 equiv) 16 1.5 99 99 3 2.5 mol% Mol. Sieves (4Å) 6 1.5 99 99 4 1.25 mol% Mol. Sieves (4Å) 6 1.5 99 99 a) Reaction conditions: 33a (0.6 mmol), Pd0-AmP-MCF, 35 (1.2 mmol), CALB (25 mg/mmol) and MS 4Å (300 mg) or Na2CO3 (0.6 mmol) were stirred in toluene (1.5 mL) under 1 atm hydrogen gas using pentadecane as internal standard. b) Determined using chiral GC and pentadecane as internal standard.

40 A DKR of 33a at 50 °C employing 5 mol% of Pd and molecular sieves (4Å) formed 34a in full conversion and 98% ee (Table 10, Entry 1). Because of the slower racemization at 50 °C compared to 70 °C the reaction required 24 h in order to reach completion. The slight decrease in ee was found to be due to a background amidation reaction, which could be avoided by changing drying agent from molecular sieves to Na2CO3 (Entry 2). Moreover, with Na2CO3 as drying agent the palladium loading could be lowered further to 2.5 mol% and still give satisfactory result; however, 36 h was required (En- try 3). The possibility to run the DKR of amines at lower temperatures enables the incorporation of other enzymes than CALB, which potentially could widen the substrate scope. This novel feature was demonstrated by employing Amano Lipase PS-C1 (lipase from Burkholdera cepacia immobilized on ceramic beads) in a DKR with 5 mol% of Pd0-AmP-MCF, affording 34a in a good isolated yield and excellent ee (Entry 4). This study constituted the first example where lipase PS was used in the DKR of amines.

Table 10. Optimization of the reaction at 50 °C and evaluation of two different lipases.a

Time Conv. ee Entry Pd-loading Additive Lipase (h) (%)b (%)b 1 5 mol% Mol. Sieves (4Å) 24 CALB 99 98 Na CO 2 5 mol% 2 3 24 CALB 99 99 (1 equiv) Na CO 3 2.5 mol% 2 3 36 CALB 99 99 (1 equiv)

Na2CO3 Lipase 4c 5 mol% 36 82d 99 (1 equiv) PS a) Typical reaction conditions: 33a (0.6 mmol), Pd0-AmP-MCF, 35 (1.2 mmol), CALB (25 mg/mmol) and MS 4Å (300 mg) or Na2CO3 (0.6 mmol) were stirred in toluene (1.5 mL) under 1 atm hydrogen gas using pentadecane as internal standard. b) Determined using chiral GC and pentadecane as internal standard. c) Reaction was carried out in dry toluene (2 mL) under 1 atm hydrogen gas using 33a (0.6 mmol), 35 (1.2 mmol), Amano Lipase PS-C1 (333 mg/mmol) and dry Na2CO3 (1 equiv). d) Isolated yield.

41 4.2.2 Scope of the reaction The scope of the DKR of amines using the nano-catalyst was investigated using the optimized conditions for 33a at 70 °C. In general, the DKR of benzylic amines proceeded smoothly, forming 34b-34d in high yields and excellent ees (Table 11, Entry 2-4). Also, bicyclic compound 34e was isolated in high yield and excellent ee (Entry 5). Substrates with an electron- donating or withdrawing group on the arene (33f and 33g) could also be transformed in high ees after a change from molecular sieves to Na2CO3 and an increase of palladium loading from 1.25 to 2.5 mol% (Entry 6-7). The change was required in order to avoid non-enzymatic amidation. Unfortunately, halogen- and nitrile-substituted arenes were not compatible with this system as they were found to undergo reduction by the Pd catalyst.

4.2.3 Recycling study One attractive feature of using a heterogeneous catalyst like Pd0-AmP-MCF is the possibility to easily separate it from the reaction mixture and reuse it. We set up a recycling study for the DKR of substrate 33a yielding 99% conversion and 99% ee in the first round (Table 12). The metal catalyst was separated from the reaction mixture, suspended in toluene and centrifuged. Excess toluene was removed with a pipette and the procedure was repeated three times before the catalyst was dried under vacuum over night. Next, the catalyst was added to a new DKR reaction, thereby starting a new cycle. The recycling was repeated four times, each with a reaction time of 15 h, before a small loss in conversion was observed. Tests of leached palladium content was possible for the first three cycles and revealed that the leaching was negligible in the first two cycles and very low in the third cycle (1.8 ppm in the reaction solvent, 0.15% of the total Pd-content). Even though the Pd content of the solvent in the last two runs was not investigated we found Pd leaching as an unlikely explanation for the loss of activity. This is supported by previous studies of Pd0-AmP-MCF under reductive conditions, where only low levels of leaching have been observed.86c,89 Instead, we reasoned that the decreased activity could be traced to losses of catalyst in the workup procedure between the runs.

42 Table 11. Scope for the DKR of amines using Pd0-Amp-MCF as racemization catalyst.a Time Yield ee Entry Substrate Product prod (h) (%) (%)

1 6 99 99

2 16 87 99

3 16 97 97

4b 16 91 98

O O HN 5 16 95 99

34e

6c 16 96 99

7c 16 89 97

a) Method A: Reaction was performed in dry toluene (1.5 mL) under 1 atm hydrogen gas using Pd0-AmP-MCF (10 mg, 1.25 mol%), 33 (0.6 mmol), 35 (1.2 mmol), CALB (25 mg/mmol) and MS 4Å (300 mg). b) 2.0 mol% Pd0-AmP-MCF used. c) Method B: Reaction was performed in dry toluene (1.5 mL) under 1 atm hydrogen gas using Pd-AmP-MCF (20 mg, 2.5 mol%), 33 (0.6 mmol), 35 (1.2 mmol), CALB (25 mg/mmol) and dry Na2CO3 (0.6 mmol).

43 Table 12. Recycling of the catalyst in the DKR.a

Cycle Conv. (%)b ee (%)c 1 99 99 2 99 99 3 98 99 4 99 99 5 90 98 a) Reaction conditions: 33a (0.6 mmol), Pd0-AmP-MCF, 35 (1.2 mmol), CALB (25 mg/mmol) and MS 4Å (300 mg) were stirred in toluene (1.5 mL) under 1 atm hydrogen gas using pentadecane as internal standard. b) Determine by GC using pentadecane as internal standard. c) Determined by chiral GC.

4.3 Conclusion The DKR of amines have been improved by the use of an efficient, recyclable catalyst consisting of Pd0 nanoparticles dispersed on a amino functionalized siliceous material (Pd0-AmP-MCF). The DKR of 1-phenylethylamine, 33a, was completed after 6 h at 70 °C using low catalytic loadings of Pd. The DKR was also performed at 50 °C; however, 24 h reaction time and an increase of catalyst were required for full conversion. Still, these result are among the best reported in terms of yield, Pd loading, reaction temperature and concentration. It was also demonstrated that the catalyst can be used for the synthesis of various substituted primary benzylic amines and also compatible with Lipase PS that previous to this report never had been used in the DKR of amines. Finally, a recycling study showed that the catalyst could be reused up to 4 times without any significant loss in activity.

44 5. Enantioselective Synthesis of Duloxetine

5.1 Introduction Even though the chemoenzymatic DKR has been investigated for a wide number of substrates, there are only a few examples where the methodology has been incorporated into a multi-step synthesis of biologically active compounds.90 More common is the reports of syntheses using hydrolases in KRs, generally leading to compounds with excellent ee.91 The benefit of upgrad- ing a synthetic pathway based on a KR into one using a DKR is the mainte- nance of the high stereoselectivity while doubling the theoretical yield. An important aspect to consider when designing an asymmetric total synthesis is the preservation of the ee throughout the synthesis. Preferably the enantiodetermining step should be late in the synthetic pathway; however, for chemoenzymatic DKR concerns have to be given to the racemization catalyst and enzyme. The enzyme has restrictions regarding the size of the substituents on the chiral center; hence it is necessary to perform the reaction on a rather small molecule, and the ability of the racemization catalyst to efficiently racemize the substrate is usually depending on the electronic properties around the chiral center. With these considerations in mind we aimed at synthesizing enantiomerically enriched Duloxetine 37, utilizing DKR on intermediate 36 to introduce the optical activity (Figure 15).

Figure 15. (R)-Duloxetine, (R)-37, can be obtained via DKR of β-hydroxy nitrile 36.

Duloxetine is a dual inhibitor acting at the reuptake of both serotonin and norepinephrine in the presynaptic cell. For patients treated with Duloxetine this results in higher concentrations of both serotonin and norepinephrine in the synaptic cleft.92 Since these neurotransmitters are intimately involved in a wide range of physiological and behavioral processes Duloxetine is pre-

45 scribed to treat symptoms of depression.93 Duloxetine, sold as a hydrochloric salt by Eli Lilly under the name CymbaltaTM, was in 2013, before the first generic version was released, the 11th most sold drug in the world with a global annual sale of $5.2 billion.94 For Duloxetine the (S)-enantiomer has been found to be twice as potent as the (R)-enantiomer and Cymbalta is therefore sold as a single enantiomer drug.93d Clearly, with such an immense economical value, many routes for the preparation of enantiopure Duloxetine are described. The vast majority of these induce enantiomeric purity by either asymmetric reduction or kinetic resolution, with the asymmetric reduction of ketone 38 and kinetic resolution through crystallization of 39 being the most common approaches (Scheme 20).95,96 The industrial process of Eli Lilly relies on the latter methodology in a resolution-racemization-recycle synthesis.97

Scheme 20. The most common methods to obtain enantiopure Duloxetine, 37.

Chiral intermediate (S)-39 is subsequently decorated with an O-naphthalene, followed by N-demethylation yielding Duloxetine (S)-37 (Scheme 21). The demethylation process requires three synthetic steps and has some critical problems, such as low yield due to byproduct formation. Thus, a synthesis from the already monomethylated, 3-(methylamino)-1-(2-thienyl)propan-1- ol (40), has been proposed as a better route for the production of Duloxe- tine.97

Scheme 21. Duloxetine, 37, can be reached in either four steps from 39 or one step from 40.

In order to prepare (R)-37 via intermediate 40, we proposed a retrosynthetic synthesis depicted in Scheme 22. We intended to gain access to enantiopure 40 by performing a DKR on β-hydroxy nitrile 36, which in turn could be synthesized from readily available and inexpensive 2-thiophene carboxaldehyde. Apart from being a precursor for molecules like Duloxetine, β-hydroxy nitriles can, through simple transformations give access to β- hydroxy acids as well as γ-amino alcohols.98 Previous reports by our group have described DKR protocols of β-hydroxy nitriles employing CALB with both Shvo’s catalyst (4) and catalyst 5.34a,98 However, prior to this study there were only a few reports on the enzymatic KR of hydroxy nitrile 36 and no DKR protocol had been reported.99 Moreover, the (S)-enantiomer of Duloxetine could potentially also be obtained from this route by performing a Mitsunobu reaction on Boc-protected intermediate 40 as previously described.99a

Scheme 22. Retrosynthetic analysis of both Duloxetine enantiomers.

47 5.2 Result and discussion The racemic key intermediate 36, was obtained through a condensation reaction between 2-thiophene carboxalydehyde and acetonitrile. The reaction was run in THF at different temperatures employing t-BuOK or LDA as bases. Using LDA at -78 °C with a one-to-one ratio of the two substrates proved to be the optimal conditions, which formed alcohol 36 in 91% yield (Scheme 23).

Scheme 23. Synthesis of β-hydroxynitrile 36.

5.2.1 Kinetic resolution As previously mentioned, the KR and racemization are often studied sepa- rately before they are combined in a DKR. In the previously described KR of compound 36, lipase PS has been used most extensively. The majority of these studies were conducted at room temperature in solvents such as DIPA and EtOAc. However, these solvents are less suitable than toluene for the transition metal-based complexes used for racemization. Hence, the enantioselectivity of different enzymes in the transesterification of 36 was measured in toluene at different temperatures employing 1 or 3 as acyl donor (Table 13). Indeed, the enzymes from Burkholderia cepacia (PS-IM, PS-C and PS-D) displayed high selectivity for the (R)-enantiomer at room temperature, but the selectivity decreased at higher temperatures (Table 13, Entry 1- 4). On the other hand, the more thermostable CALB was capable of main- taining high selectivity even at elevated temperatures (entry 5-8). Two (S)- selective enzymes, Subtilisin Carlsberg and CALB mutant W104A, developed by Hult and co-workers, 100,101 were also investigated but both failed to distinguish between the enantiomers of 36 (Entry 9-11).

48 Table 13. Investigation of the selectivity of different enzymes in the kinetic resolution of β-hydroxynitrile 36.a

Temp Time Conv.b ee c ee c Entry Enzyme OH OAc Ed (ºC) (h) (%) (%) (%) 1 PS-IM rt 3 35 52 97 110 (R) 2 PS-C rt 3 44 76 96 112 (R) 3 PS-D rt 2 33 54 98 170 (R) 4 PS-D 80 2 38 51 82 16 (R) 5 CALB rt 6 6 6 >99 >200 (R) 6 CALB 50 4 21 25 >99 >200 (R) 7 CALB 70 2 23 27 98 128 (R) 8 CALB 80 4 35 56 96 86 (R) 9 CALB-W104Ae 50 24 14 14 81 10 (S) 10f Sub. Carlsbergg 38 4 41 5 8 1.2 (S) 11f Sub. Carlsbergg 50 5 23 6 18 1.5 (S) a) Reaction conditions: 36 (0.25 mmol), enzyme (50 mg/mmol), Na2CO3 (0.25 mmol), isopropenyl acetate (1) (2 equiv) stirred in dry toluene (0.5 mL). b) Measured by 1H NMR. c) Measured by chiral HPLC. d) Values calculated from eeOH and eeOAc accodring to Eq (3), the selectivity displayed by the enzyme is given in parenthesis. e) A CALB variant proven to be 101 (S)-selective. f) Reaction conditions: 36 (0.1 mmol), subtilisin (100 mg/mmol), Na2CO3 (0.1 mmol), 2,2,2-trifluoroethyl butyrate 3 (3 equiv) in toluene (0.2 mL) g) 1:4:4 subtilisin:Brij:Octyl-β-D-glucopyranoside.70

5.2.2 Racemization Earlier racemization studies on β-hydroxy nitriles have revealed that elevated temperatures are often required for these alcohols to be efficiently racemized. This is due to the electron-withdrawing nitrile disfavoring the hydride abstraction.102 The racemization was studied using activated monomeric complex 5 at room temperature, 50, 70 and 80 °C (Table 14). At 70 °C and above the enantiomerically pure alcohol 36 was racemized within 5 minutes (Entry 1-2), whereas it took 20 minutes to achieve the same result at 50 °C (Entry 3). At room temperature alcohol 36 still had an ee of 57% after 2 h reaction time confirming that the thiophene nitrile is more challenging to racemize than analogues benzylic sec-alcohols with alkyl substituents (Entry 4).

49 Table 14. Racemization of (S)-β-hydroxy nitrile 36 at rt, 50, 70 and 80 °C.a

Temp Time eeb Entry (°C) (min) (%) 1 80 5 0 2 70 5 0 3 20 20 0 4 rt 120 57 a) Reaction condidions: 36 (0.40 mmol), Ru-cat. 5 (5 mol%), t-BuOK (6 mol%), Na2CO3 (0.40 mmol) were stirred in dry toluene (0.8 mL) b) Measured by chiral HPLC.

5.2.3 DKR The racemization study made it clear that an elevated temperature would be required for the DKR; hence a reaction was set up at 70 °C. However, even though both the racemization and enzymatic reactions were expected to proceed well at this temperature the result was unsatisfactory and delivered the product in only a moderate ee of 81%. Furthermore, under these conditions, the fully conjugated elimination product 42 was formed in 20% yield (Scheme 24).

Scheme 24. Unsuccessful DKR of rac-36 giving rise to a considerable amount of byproduct 42.

In order to elucidate why the resulting ee did not correspond to what was expected from the E value, a control experiment was performed. The reaction was performed under DKR conditions, omitting the enzyme; showing that a non-enzymatic acetylation was taking place as a background reaction. The background reaction would be disfavored and hopefully avoided if the reaction was performed at a lower temperature and to test this hypothesis a DKR was carried out at 50 °C. Gratifyingly, the ee of the acetate was increased to 94% and simultaneously the amount of elimination product was lowered to 6% (Scheme 25). However the reaction was much slower at this temperature and required 48 h to reach 81% yield of the product.

Scheme 25. Result from the DKR of rac-36 at 50 °C.

The rate can by enhanced by increasing the amount of CALB and/or acyl donor. Hence, the potential effect of these adjustments was investigated by three kinetic resolutions of 36 at 50 °C (Table 15). By doubling the enzyme loading the yield of acetate (R)-41 was increased from 21 to 33%, while keeping a high enantioselectivity (Table 15, Entry 1-2). Increasing the amount of acyl donor from 2 to 3 equiv had no significant effect and the same stereoselectivity within measurement margin of error was observed (Table 15, Entry 3).

Table 15. Optimization of KR in order to find suitable reaction conditions for the DKR.a

CALB 1 Conv.b ee c ee c Entry OH OAc E (mg/mmol) (equiv) (%) (%) (%) 1 50 2 21 25 >99 >200 2 100 2 33 47 98 158 3 100 3 32 44 98 152 a) Reaction conditions: 36 (0.5 mmol), Na2CO3 (0.5 mmol), isopropenyl acetate (1) (2 equiv) were stirred in dry toluene (1 mL). b) Measured by 1H-NMR. c) Measured by chiral HPLC.

By performing the DKR reactions with small amounts of water it was found that the amount of elimination product increased. In the presence of water, the acyl donor and the acetate (R)-41 can possibly be hydrolyzed forming acetic acid. If the heterogeneous Na2CO3 does not neutralize the acid immediately, these slightly acidic conditions might promote protonation of the product, which in turn initiates elimination. Taking extra precaution in the drying of all reaction components and also employing higher concentrations of acyl donor and enzyme (3 equivalents and 100 mg/mmol respectively) a DKR set up at 50 °C gave a positive outcome. The reaction yielded minimal amount of the elimination product (2%) and reduced the effect of the non-

51 enzymatic esterification, making it possible to isolate the acetate in 87% yield and in 98% ee (Scheme 26).

Scheme 26. Optimized conditions and best results obtained for the DKR of 36.

5.2.4 Total synthesis of Duloxetine With the enantiomerically pure (R)-41 at hand the subsequent steps towards intermediate 40 and eventually Duloxetine (37) could be conducted (Scheme 28). Reduction of the nitrile and acetate of 41 was first tried with 3 equiv of LiAlH4; however, the desired product 43 was not formed. Instead large amounts of decomposition products were observed. We therefore turned to a described procedure using borane dimethylsulfide (BH3•SMe2) in THF, ac- complishing the formation of amino alcohol 43 (Scheme 27).103

Scheme 27. Reduction of compound 41 followed by addition of ethyl chloroformate in order to yield N-carbamate protected compound 44.

In order to avoid the tedious workup associated with amino alcohols the crude reaction mixture was only treated with MeOH to quench any excess hydride reagent before being transferred to a two-phase system of aqueous NaHCO3 and DCM containing ethyl chloroformate. This afforded carbamate 44 in 67% over two steps (Scheme 27 and 28). N-methyl compound 40 was obtained in 87% yield after treating 44 with an excess of LiAlH4 in refluxing THF over night (Scheme 28).

52 In the last step, a nucleophilic aromatic substitution of 40, employing 1- fluoronaphthalene as electrophile in DMSO would form Duloxetine. Accord- ing to literature procedures the reaction should start by deprotonation of the substrate with 1 equiv of NaH at 50 °C for 1h, followed by addition of 1- fluoronaphthalene at 95°C.96d In our hands this method resulted in a severe racemization of the product. When less harsh conditions were used, deproto- nating at room temperature and adding the electrophile at 50 °C, the reaction did not go to completion. However, the enantiomeric excess was retained in the product. To increase the yield the amount of NaH was increased to 1.5 equiv. This led to a drop of ee, which showed that only 1 equiv of the base was to be added. To ensure that the base was not consumed by anything else but alcohol 40 the reaction was carried out under argon atmosphere with freshly distilled and degassed DMSO. After addition of NaH, the mixture was stirred vigorously to assure full solvation of the base before 40 was added. The substrate was stirred with the base for 30 minutes at room temperature before 1-fluoronaphthalene was added, and the reaction temperature was increased to 50 °C. After an additional hour, (R)-Duloxetine ((R)-37) was obtained in 80% yield, with almost complete retention of stereochemistry.

Scheme 28. Synthesis of (R)- and (S)-Duloxetine.

53 5.3 Conclusion A total synthesis of (R)-Duloxetine, (R)-37, has been developed utilizing DKR in the enantiodeterming step. An overall yield of 37% was obtained over six steps starting from an inexpensive and simple aldehyde precursor (Scheme 28). The DKR proceeded smoothly, delivering key intermediate 36 in high yield (87%) and in high enantiomeric excess (98%). The synthetic route also constitute a formal total synthesis of (S)-Duloxetine.

54 6. Investigation of the Impact of Water on the Enantioselectivity Displayed by CALB in the Kinetic Resolution of δFunctionalized Alkan 2ol Derivatives

6.1 Introduction Because of its high stereoselectivity and activity, CALB is one of the most widely used lipases in KRs and DKRs of sec-alcohols, and its properties and catalytic machinery have been studied intensively.104 CALB’s active site is accessed through a funnel-shape structure, creating a narrow binding site in which the large-sized group of the alcohol and the acyl chain of the ester moiety are oriented during catalysis. The catalytic triad of the active site is built up by Ser105, His224 and Asp187, and the negative charge of the tetrahedral intermediate (TI) is stabilized by the hydroxyl group from the side chain of Thr40 and the two back-bone amides of Gln106 and Thr40. The active site also comprises a small cavity called the stereospecificity pocket that hosts the medium-sized group of the fast reacting enantiomer of a sec- alcohol during catalysis (Figure 16). 105

Thr 40 Ile 285

Ser 105 Ala 281

Leu 278

Asp 187 Trp 104 His 224 Figure 16. Active site of CALB with the serine bound tetrahedral intermediate of 2- hexanol. The large pocket, shown in red, points towards the entrance of the active site. The stereospecificity pocket, shown in green, accommodates the α-methyl group. The negative charge is stabilized in the oxyanion hole (dashed lines from Thr40).

55 The slow-reacting enantiomer cannot access the active site without orienting its large group into this small stereospecificity pocket, leading to a low specificity for this enantiomer. Hence, for compounds that have a clear size difference between the large and medium groups, CALB generally shows excellent enantioselectivity with minimal to no reaction of the slow-reacting enantiomer.106 Our group found an exception to this general trend in a study where un- symmetrical 1,4-diols like rac-45 was converted into γ-hydroxy ketone (R)- 48 using a combined DYKAT and oxidation protocol (Scheme 29). The hydroxyl group in the least sterically hindered position was acylated by CALB, whereas the other hydroxyl group was not due to the surrounding steric bulk. Catalyst 4, employed for racemization, had a dual effect and oxidized the hindered alcohol into a ketone, forming the desired γ-hydroxy ketone. De- spite CALB showing a pseudo E value107 of 285 for rac-45 (corresponding to a theoretical ee in DKR of >99%) the ee of (R)-48 was only 90%. The cause of this drop from the theoretical ee was found to originate from a very poor E value of only 2.2 for the intermediate δ-keto alkan-2-ol 47 (corresponding to a theoretical ee in DKR of 38%). Thus, a highly enantioselective reaction would only be obtained if rac-45 was first acetylated to (S,R)/(R,R)- 46, then oxidized to (R)-48, while the reversed reaction order first forming oxidized rac-47, would result in a low enantioselectivty (Scheme 29). Since both the selective and unselective process occurred simultaneously, the product ee was lower than first expected.

Scheme 29. Example of DYKAT forming γ-hydroxy ketone (R)-48 from unsymmet- rical 1,4-diol rac-45. Enantiomeric ratio of the two CALB catalyzed processes given.

A similar finding was made in a project converting symmetrical 1,4-diol 49 into 1,4-diacetate 51.108 Acetylation of the first hydroxyl group proceeded with high enantioselectivity; however, upon acetylation of monoester 50, a

56 lower selectivity was observed. The selectivity drop was found to depend on the relative configuration of the acetoxy group and the alcohol. For anti-5- acetoxy-2-hexanol the E value was as low as 7, whereas its syn disasteromer had an E value of 94.

Scheme 30. Stereoselectivities observed in the CALB catalyzed transesterification of a) meso-49 and b) rac-49 in toluene.

To gain a deeper understanding of the low enantioselectivity for CALB- catalyzed transesterification reactions of alkan-2-ols bearing a δ-keto or δ- acetoxy substituent, a computational study by molecular dynamic simulations was performed in our group. Simulations of the substrate-enzyme TIs of both unfunctionalized and δ-hydroxy, δ-acetoxy and δ-keto functionalized alkan-2-ols were studied.109 The simulations showed that the stereoisomers of diols and unsubstituted alkan-2-ols, where the acylated chiral center was (R)-configured, adopted a relaxed conformation with the medium-sized group oriented towards the stereospecificity pocket. The stereoisomers having an (S)-configured chiral center had a strained, less favorable conformation. This is in accordance with the high E values that CALB displays for these substrates.110 δ-Acetoxy alkan-2-ols, did not fit as well as the 1,4 diols, and the acetoxy group in the δ-position had to be positioned in a small polar cavity available in the large binding pocket formed in the gap between Ala281 and Ala282 (Figure 17). In this pocket, acetates on an (R)-configured chiral center had a

57 more favorable conformation than the (S)-configured epimer, resulting in CALB favoring transesterification of monoacetates having the δ-acetate chiral center of (R)-configuration.

a) b)

Ala 282

Thr 40 Thr 40

Figure 17. Serine bound tetrahedral intermediates formed in the acetylation of a) (R,R)-50 (green) orienting the OAc group towards Ala282. This conformation is not accessible for (S,S)-50 (blue); b) (R)-52 (green) and (S)-52 (blue) that both adopts binding mode without enantiorecognition.

For δ-keto alkan-2-ols, a strong electrostatic repulsion was observed between the two partially negatively charged carbonyls on the substrate and amino acid Thr40. Both enantiomers of 5-hydroxyhexan-2-one, 52, had to change their position directing the carbonyl group away from Thr40, consequently forcing the substrate to elevate from the active site. This led to that the α- methyl group of both enantiomers was forced out of the stereo-specificity pocket, explaining the low enantiodiscrimination (Figure 17b). Interestingly, when a water molecule was positioned close to Thr40 in the model, a hydrogen bond was established from the water molecule to the δ- keto substituent. This allowed the substrate to be positioned deeper into the active site, making the small group of the (R)-enantiomer fit favorably in the stereospecificity pocket (Figure 18a). However, this water effect was never tested experimentally.

a) b)

Thr 40 Thr 40 His 224 His 224

water Ser 105 Ser 105 Ser 281

Trp 104 Trp 104 Figure 18. Position of the serine bound TI of (R)-52 (green) and (S)-52 (blue) in: a) wild type CALB with a water molecule placed next to Thr40; b) CALB mutant Ala281Ser. The (R)-enantiomer is in both cases able to orient its medium-sized group into the stereospecificity pocket due to the hydrogen bond between the carbonyl oxygen of (R)-52 and the water molecule or Ser281.

58 Based on the modeling, a plausible enzyme variant capable of mimicking this water molecule was suggested. By exchanging the alanine residue at position 281 to a serine residue, a hydrogen bonding hydroxyl group was installed at approximately the same site as the water molecule (Figure 18b). The variant was produced, and evaluated in one single experiment where it displayed high enantioselectivity for 52. However, only this substrate was investigated and CALB variant was not investigated for any of the δ-acetoxy substrates. By performing reactions with both wild type CALB and CALB variant A281S, we intended to test the hypothesis of the hydrogen bond aiding CALB’s ability to perform an enantioselective reaction for δ-keto alkan-2- ols. We were also interested to see if a hydrogen bond effect is observed for anti-δ-acetoxy alkan-2-ols.

6.2 Results

6.2.1 Transesterification vs. hydrolysis In theory, the hypothesis of the water molecule close to Thr40 would imply that a transesterification reaction performed in an aqueous solution instead of a dry organic solvent potentially could show increased stereoselectivity. Experimentally though, performing the transesterification in an aqueous solution is not feasible since the presence of water will induce hydrolysis of the formed ester. However, by realizing that the hydrolysis and transesterification both pass through the same TI, the theory can be tested simply by performing the hydrolysis reaction of the δ-functionalized alkan- 2-ols (Figure 19). If enantioselectivity arises in these reactions, a strong support for this hypothesis is acquired.

Figure 19. Illustrating the formation and deformation of the shared tetrahedral intermediate of the hydrolysis and transesterification

59 Transesterifications of the δ-acetoxy and δ-keto functionalized alcohols 46- 47 and 52-54 were performed in toluene with isopropenyl acetate (1) and immobilized CALB in the form of Novozym-435 (marked by Novozymes). As predicted, δ-keto substituted alcohols rac-47, rac-52 and rac-53 all had low E values of 10 or below (Entry 1-3). The stereoselectivity for anti- acetoxy substrates rac-46 and rac-54 was modestly better with E values of 18 and 36, respectively (Table 1, Entry 4-5). Comparing these results with that obtained in the transesterification of unsubstituted alkan-2-ol 55 (Entry 6), it is clear that the δ-group has a negative effect on the enantioselectivity.

Table 16. CALB catalyzed transesterification and hydrolysis of various alkan-2-ols. Transesterificationa Hydrolysisb Entry Substrate Ec Entry Substrate Ec

1 6 (3)d 7 190

2 9 (45)e 8 170

3 3 9 431

4 18 10 >500

5 36 11 331

6 >200 12 >200 a) Reaction conditions: rac. sec-alcohol (0.1 mmol), supported CALB (Novozym 435, 40 mg/mmol), isopropenyl acetate (1) (2 equiv), Na2CO3 (0.1 mmol) stirred in dry toluene (0.2 mL) at 30 °C. b) Reaction conditions: rac-acetate (0.015 mmol), CALB solution (Lipozyme, 4 µL) in KPO4 buffer (50mM, pH 7.0)/MeCN (85:15) shaken with 290 rpm at 30 °C. c) For conversions and ees see Paper VI. d) Lipozyme on celite e) Reaction performed in toluene saturated with water (aw=1)

60 In accordance with the previous prediction, introduction of water would result in an increase of specificity for the (R)-enantiomer. This was first tested by performing the transesterification of 52 in wet toluene, taking a sample rapidly before the hydrolysis reaction was significant. The wet solvent increased the E value of CALB to 45 (Table 16, Entry 2). The hydrolysis of 48 and 56-60 was performed in aqueous phosphate buffer solution using a liquid preparation of CALB, “Lipozyme”, also acquired from Novozymes. In all cases the reaction proceeded with excellent enantioselectivity, strongly indicating that CALB is highly stereoselective for these substrates if water is present in the active site (Table 16, 7-12). In order to exclude that the difference in enantioselectivity originate from the fact that two different CALB sources had been used (liquid preparation and immobilized on polyacrylic resin) the transesterification of substrate 47 was repeated with Lipozyme immobilized on celite. However, a low E value was still observed (Table 16, Entry 1).

6.2.2 CALB A281S The influence of the hydrogen bond can be further supported by mutating Ala281 to a Ser (CALB A281S), thereby creating an enzyme capable of mimicking the proposed water molecule around Thr40. The mutant was produced by site directed mutagenesis and expressed in P. pastoris. After cen- trifugation, the supernatant containing the excreted CALB A281S was con- centrated to reach a concentration of 1 g enzyme/mL buffer and immobilized onto celite in order to increase the stability of the enzyme in organic solvents. The enantioselectivties for substrate 46-47 and 52-54 was tested in a transesterification reaction for 24 h using 1 as acyl donor. For δ-keto alkan- 2-ols CALB A281S was active and displayed enantioselectivities, three to twelve times higher than the wild type. The δ-acetates 46 and 54 were not acetylated by the enzyme, the reason was not elucidated, but we hypothesiz- es that the change from alanine to serine reduced the size of the small pocket accommodating the δ-acetoxy group, inhibiting 46 and 54 from proper binding.

61 Table 17. Transesterification catalyzed by CALB A281S and wild type CALB.a acyl donor 1 (2 equiv) R CALB or CALB A281S R R OH Na2CO3,toluene30°C OAc OH Eb Eb Selectivity Entry Substrate Product CALB wt increase A281S CALB (%)

1 19 6 317

2 115 9 1278

3 19 3 633

4 nr. 18 -

5 nr. 36 - a) Reaction conditions: (i) for CALB A281S; rac. sec-alcohol (0.1 mmol), supported CALB A281S (on Celite, 40 mg/mmol), isopropenyl acetate (1) (2 equiv), Na2CO3 (0.1 mmol) in dry toluene (0.2 ml), performed at 30 °C, reaction time 24 h. (ii) for wt CALB; racemic sec- alcohol (0.2 mmol), supported CALB (Novozyme 435, 40 mg/mmol), isopropenyl acetate (1) (2 equiv), Na2CO3 (0.2 mmol) in toluene (0.2 ml), performed at 30 °C. b) Calculated according to Eq (1), (2) or (3). For conversions and ees see Paper VI.

The enantioselectivity for CALB A281S-catalyzed hydrolysis of rac-56 was determined to 118 (Scheme 31). This is a lower E value than that obtained with the wild type; however, it is, within experimental error, identical to the E value that CALB A281S displayed in the transesterification of rac-52. This indicates that the serine-substrate interaction is present in the active site rather than the water-substrate interaction and that the mutation is responsi- ble for the enantioselectivity increase in the ester formation.

Scheme 31. Hydrolysis of rac-56 catalyzed by CALB variant A281S.

62 6.3 Discussion and Conclusion CALB catalyzes the hydrolysis and formation of δ-functionalized alkan-2- acetates with very different enantioselectivites. Based on an earlier publication that analyzed the enantioselectivity of CALB with molecular dynamics, we hypothesize that the ability of CALB to perform an enantioselective reaction of these substrates is dependent on the presence of a hydrogen bonding entity, close to Thr40 in the active site. The simulation predicted that the hydrolysis reaction, having a hydrogen bonding water molecule in the active site, would allow the (R)-enantiomer to adopt a beneficial conformation, whereas the (S)-enantiomer would be tilted away from the stereospecificity pocket. In the transesterification both enantiomers would adopt a non- discriminating bonding, leading to a reaction with low stereoselectivity. For δ-keto alkan-2-ols both the result obtained in the hydrolysis and transesterification, in combination with that of the mutated version of CALB, mimicking the presence of this water molecule, supports the theory regarding the hydrogen bond stabilizing effect with the (R)-enantiomer, resulting in a stereoselective reaction. For the δ-acetoxy alkan-2-ols the result is not as clear. The hydrolysis proceeds in high stereoselectivity, supporting the hydrogen bonding theory, however no computational modeling has been made for this substrate class and the mutant lacks activity for the substrate. Further experiments are needed before any conclusion can be made. It is worth mentioning that water previously has been shown to effect the enantioselectivity of enzyme catalyzed reactions. Alteration of the water activity has resulted in increased, decreased and unaffected enantioselectivities. The differences are most often explained by the ability of waters to act as a lubricant increasing the amino acid mobility, increasing the active site polarity and increasing hydrophobic interactions between the enzyme and substrate.111113 A report by Graber and co-workers goes deeper on the effect of water on the enatioselectivity displayed by CALB.114 By increasing the water activity from 0.02 to 0.2 in a transesterification reaction performed on pentan-2-ol in a solid/gas reactor they could see that enantioselectivity increased from 101 to 320. With the aid of molecular modeling they proposed that water could fill up parts of the stereospecificity pocket, making it smaller, hence the enantiodiscrimination was larger. It is not unlikely that the effect observed in this project might be influenced by this factor as well.

63 7. General conclusions and outlook

This thesis focuses on the investigation and improvement of enzyme- catalyzed reactions. The major part of the projects involves DKR for various types of substrates. Racemization of alcohols and amines were achieved using ruthenium-based complexes and nano Pd-particles, respectively. In general, high yields and excellent ees were obtained under mild reaction conditions. In the first two projects N-hetereocyclic alcohols and cyclic allylic alcohols were transformed into their corresponding enantioenriched esters at temperatures between 50 or 60 °C. The results constitute an efficient way of preparing these compounds with high enantioselectivity. These studies also provided an increased knowledge about enzyme selectivity and the function of the racemization catalyst, which could be valuable in future developments of DKR protocols. In the third project the DKR of benzylic amines was improved by utilizing a recyclable heterogeneous Pd nanoparticle catalyst. It was previously known that the Pd catalyst was capable of efficiently racemizing amines; however, we optimized its performance in DKR. The reaction works at higher substrate concentrations, requiring lower catalyst loadings and reaction temperatures than most other DKRs. Moreover, we completed the enantioselective synthesis of (R)-Duloxetine, introducing the optical activity by a DKR of a β-hydroxy nitrile. The synthetic pathway started from inexpensive 2-thiophenecarboxaldehyde, and the target structure was reached in six steps with an overall yield of 37% and ee above 96%. The last project was a study of CALB, one of the most widely used transesterification enzymes in organic synthesis. A theory based on molecular modeling simulations postulated that a water molecule in the active site of CALB potentially could have a positive effect on the enantioselectivity of certain δ-functionalized alkan-2-ols. The enantioselectivities that wild type CALB displayed in the aqueous hydrolysis reactions were superior to those obtained in the reverse transesterification reactions in dry organic solvent. This finding, together with the increased enantioselectivty observed for CALB variant A281S in transesterification of δ-functionalized alcohols, supported the postulated theory.

64 Appendix A

The author’s contribution to publications I-VI:

I. Designed the project in collaboration with RM. Performed the major part of the synthetic work. Wrote half of the article.

II. Designed the project in collaboration with RM. Performed the majority of the synthetic work. Wrote the article.

III. Designed the project in collaboration with JR. Performed the synthetic work. Wrote half of the article.

IV. Designed the project in collaboration with KG. Performed parts of the experimental work. Took part in writing the article.

V. Performed the majority of the experimental work under supervision from AT. Took minor part in writing the article

VI. Designed the project in collaboration with BY. Performed a minor part of the experimental work. Wrote half of the article.

65 Appendix B

Reprint permissions were kindly granted for each publications by the following publishers:

66 Acknowledgements

I would like to express my appreciation to:

My supervisor, Prof. Jan-Erling Bäckvall for accepting me as a Ph. D. stu- dent in his group, for guidance, for love of science, and for nice conferences and dinners throughout the years!

Dr. Annika Träff for introducing me to the world of DKR and for guiding me through my first research project. Since you left me I have on several occasions missed you and your chemistry, planning and dancing skills!

Dr. Renaud Millet. What would I have done without you? You were a good mentor and, for me, the perfect collaboration partner! The projects we performed together were filled with joy!

Mr. Karl Gustafson, Dr. Oscar Verho and Dr. Karin Engström for a job well executed! Thanks to you guys the world can now perform DKR of benzylic amines at only 50 °C. Yippie! A golden star to Karl for a smooth, productive and fun collaboration!

Dr. Jerk Rönnols and Prof. Göran Widmalm for a truly nice and well needed collaboration between our groups. Jerk, I enjoyed watching you doing NMR things on the computer. It felt like I was 10 years old again, watching Martin play Super Mario Bros. 2.

Bin Yang a.k.a. “Mike Tommasi” for the hard work you put into our project. Your stubbornness and determination got us through! Keep on struggling and soon you will be able to write your own acknowledgement.

Christopher Lood, Renaud Millet, Andreas Persson, Moa Lihammar, Oscar Verho, Karl Gustafson and Karim Engelmark Cassimjee who helps me found erors and improve the text quality at this thesis!

67 My highly efficient, well-organized and punctual German Erasmus students Marius Hartmer, Sabrina Molitor, Eric Seidel and Tobias Schnitzer. You were all very special and I’m happy that I could teach you things about both chemistry and life.

Travel grants from Ångpanneföreningens Forskningsstiftelse scholarship, K & A Wallenberg’s travel scholarship and C&C L foundation which allowed me to travel to conferences. I would also like to thank those colleagues that kept me company at the lectures, beaches, vineyards and vulcanos.

CALB and catalyst 5 for being efficient catalysts, even at room temperature, and for being able to convert racemic alcohols into enantioenriched esters in high yields and excellent ees.

The TA staff: Martin, Christina, Sigrid, Ola, Carin, Kristina, Louise, Jenny and Britt for keeping the department together and for your nice greetings every morning when I enter the kitchen!

Past and present members of the Jeb group who created a nice research environment and gave me immense help and support on several occasions: Maria K., Xu, Jan, Tamás, Karim, Lisa, Dr. Deng, Mozaffar, Anders, Renaud, An- nika, Antoine, Sascha, Eric, Yoshi, Jonas, Karin, Teresa, Javier, Tanja, Beneesh, Andreas, Min, Ylva, Santosh, Chicco, Madde, Maria H., Markus, Erik, Nicolas, Gillian, Laura, Nerea, Oscar, Mouli, Can, Debasis, Tuo, Anuja, Bin, Karl, Johanna, Christoph and the ones I might have missed.

My fume-hood neighbors who taught me many things and filled my days with joy: Tuo Jiang a.k.a. “Jay Brucesome”. You are a very nice and strange man who possesses many unique skills, such as the ability to shoot down airplanes with the power of your mind. Keep on being retarded! Michiel Petrus. Even if your stay next to me was short your photo never left my fume-hood. No matter how stökigt my fume-hood was! Chicco “Xisco” Manzuna Sapo. You really lifted mine and the group’s spirit during your stay at SU. I don’t know how, but you did it! Yeah!

Andreas, for the support and advice you have given me! I’m sorry I didn’t listen to all of them…

Teresa for endless chats about everything and everyone! You are a wonderful enigma that I will never fully understand.

68 Ylva for nice trips, wine, spexs, laughter, lay-out and biochemistry! My Ph.D. years would not have been as good without you!

All the past and present people at the department. It has been an honor to get to know all of you! Special thanks to Elias and the GW guys Robert, Jerk, Thibault and Christopher “som paxat allt” for nice lunches, fikas, parties etc.!

Prof. Ulf Nilsson, Prof. Ulf Ellervik and Prof. Em. Torbjörn Frejd for being brilliant lecturers and supervisors! I especially would like to thank Ulf Ellervik for inspiring me to continue with organic chemistry and introducing me to the world of chemistry shows!

My super cool organic chemistry study partners: Christopher Lood whom I would be nothing without, Martin Axelsson for exploring the world with me and Olle Engström for boardgames, EU3 and non-chemical conversations.

All friends outside the department whom I’ve double booked on several occasions! I’m grateful to Henrik, Tove, Mats, Katarina, Joel and Petra for being awesome and allowing me to be weird. Calle, Moa, Johannes and Karro for nice dinners and trips. Frida and Jessica for support and talks!

Jerker, Birgitta, Oskar, Hanna, Östen and Gunnel for being the best extra family I could ever have wished for!

Sussie, Elin, Martin and Tomas. I (now) love you all equally much and I’m very happy for growing up with such weird people as you. Also Elliot, Mårten, Ella, Julia, Emma and Elin går inte av för hackor!

Peo & Cilla for taking care of me like you created me. Thank you!

Mamma & Pappa for creating me and for your unconditional love and support!

Last but not least my wonderful wife Moa for endless support, love and understanding! I’m so happy to have you in my life and that you keep me company in good times and bad. After becoming a Ph.D. I might even get the courage to kiss you in public. Love you!

69 Reference

[1] Following ACS Standard Abbreviation/Acronyms, 2014 Guidelines for Au- thors, Organic Letters Home Page. http://pubs.acs.org/page/orlef7/submission/authors.html (access Oct 2014) [2] Wolf, C. Dynamic Stereochemistry of Chiral Compounds: Principles and Applications; The Royal Society of Chemistry: Cambridge, 2008; pp 71–75. [3] Patocka, J.; Dvorak, A. J. Appl. Biomed. 2004, 2, 95–100. [4] McConathy, J.; Owens, M. J. Prim. Care Companion J. Clin Psychiatry 2003, 5, 70–73. [5] Fassihi, A. R. Int. J. Pharm. 1993, 92, 1–14. [6] Anderson, R. J.; Groundwater, P. W.; Todd, A.; Worsley, A J. Antibacterial Agents: Chemistry, Mode of Action, Mechanisms of Resistance and Clinical Applications; John Wiley and Sons: Chichester, 2012; pp 231–242. [7] a) Breuer, M.; Ditrich, K.; Habicher, T.; Hauer, B.; Kesseler, M.; Stuermer, R.; Zelinski, T. Angew. Chem. Int. Ed. 2004, 43, 788–824. b) Farina, V.; Reeves, J. T.; Senanayake, C. H.; Song, J. J. Chem. Rev. 2006, 106, 2734– 2793. [8] Olbe, L.; Carlsson, E.; Lindberg, P. Nat Rev Drug Discov 2003, 2, 132–139. [9] Gawley R. E.; Aubé. J. Principles of Asymmetric Synthesis; Elsevier Science Ltd: Oxford, 1996; pp 1–43. [10] Drauz, K.; Waldmann, H. Enzyme Catalysis in Organic synthesis: A Com- prehensive Handbook; Wiley-VCH: Weinheim, 1995; pp 1–32. [11] Buchner, E. Ber. Dt. Chem. Ges. 1897, 30, 1110–1113. [12] García-Urdiales, E.; Alfonso, I.; Gotor, V. Chem. Rev. 2011, 111, 110–180. [13] Bornscheuer, U. T.; Kazlauskas, R. J. Hydrolases in Organic Synthesis: Re- gio- and Stereoselective Biotransformations; Wiley-VCH: Weinheim, 1999; pp 1–3 and 39–64. [14] With the minor exception of the polish scientist Ernest Alexander Sym: Sym E. A. Biochem. Z. 1933, 258, 304–324. Commented in: Halling, P. J.; Kvittingen, L. Trends Biotechnol. 1999, 17, 343–344. [15] a) Summarized in: Carrea, G.; Riva, S. Organic synthesis with enzymes in non-aqueous media; Wiley-VCH: Wennheim, 2008. First examples: b) Zaks, A.; Klibanov, A. M. Science 1984, 224, 1249–1251. c) A.; Klibanov, A. M. Proc. Natl. Acad. Sci. U. S. A., 1985, 82, 3192–3196.

[16] Halling, P. J. Phil. Trans. R. Soc. Lond. B 2004, 359, 1287–1297. [17] a) Kraut, J. Annu. Rev. Biochem. 1977, 46, 331–358. b) Bousquet-Dubouch, M.-P.; Graber, M.; Sousa, N.; Lamare, S.; Legoy M.D. Biochimica et Biophysica Acta, 2001, 1550, 90–99. [18] Brady, L.; Brzozowski, A. M.; Derewenda, Z. S.; Dodson, E.; Dodson, G.; Tolley, S.; Turkenburg, J. P.; Christiansen, L.; Huge-Jensen, B.; Norskov, L.; Thim, L.; Menge, U. Nature 1990, 343, 767–770. [19] Warshel, A.; Natay-Szabo, G.; Sussman, F.; Hwang, J.-K. Biochemistry 1989, 28, 3629–3637. [20] Kazlauskas, R. J.; Weissflock, A. N. E.; Rappaport, A. T.; Cuccia, L. A. J. Org. Chem. 1991, 56, 2656–2665. [21] Chen, C.-S.; Fujimoto, Y.; Girdaukas, G.; Sih, C. J. J. Am. Chem. Soc. 1982, 104, 7294–7299. [22] Faber, K. Biotransforamtion in Organic Chemistry; Springer-Verlag: Berlin, 2011; pp 38–39. [23] Steinreiber, J.; Faber, K.; Griengl, H. Chem. Eur. J. 2008, 14, 8060–8072. [24] Ebbers, E. J.; Ariaams, G. J. A.; Houbiers, J. P. M.; Bruggink, A.; Zwanen- burg, B. Tetrahedron 1997, 53, 9417–9476. [25] a) Troch-Grimshaw, J.; Henbest, H. B. Chem. Commun. 1967, 544–544. b) Chowdhury, R. L.; Bäckvall, J.-E. J. Chem. Soc., Chem Commun. 1991, 16, 1063–1064. c) Gao, J.-X.; Ikariya, T.; Noyori, R. Organometallics 1996, 15, 1087–1089. d) Bäckvall, J.-E. J. Organometal. Chem. 2002, 652, 105–111. e) Samec, J. S.; Bäckvall, J.-E.; Andersson, P. G.; Brandt, P. Chem. Soc. Rev. 2006, 35, 237–248. f) Peris, E.; Crabtree, R. H. Iridium N-Heterocyclic Carbene Complexes and Their Application as Homogeneous Catalysts. In Iridium Complexes in Organic Synthesis; Oro, L. A.; Claver, C., Eds.; Wiley- VCH: Weinheim, 2008; pp 39–54. g) Saidi, O.; Williams, J. M. J. Top. Organomet. Chem. 2011, 34, 77–106. [26] Faber K. Chem. Eur. J. 2001, 7, 5004–5010. [27] a) Pàmies, O.; Bäckvall, J.-E. Chem. Rev. 2003, 103, 3247–3262. b) Ahn, Y.; Ko, S.-B.; Kim, M.-J; Park, J. Coord. Chem. Rev. 2008, 252, 647–658. c) Lee, J. H.; Han, K.; Kim, M. J.; Park, J. Eur. J. Org. Chem. 2010, 6, 999– 1015. [28] a) Verzijl, G. K. M.; De Vries, J. G.; Broxterman, Q. B.; (DSM N. V., Neth- erlands). Process for the preparation of enantiomerically enriched esters and alcohols. Application: WO2001090396 A1, 2001. b) Broxterman, Q. B.; Verzijl, G. K. M.; (Merck & Co., Inc., USA). Process for the synthesis of (R)-1(3,5-bis(trifluoromethyl)phenyl)ethan-1-ol and esters thereof by dynamic kinetic resolution. Application: WO2003043575 A2, 2003. [29] Dinh. P. M.; Howarth, J. A.; Hudnott, A. R.; Williams, J. M. J.; Harris, W. Tetrahedron Lett. 1996, 37, 7623–7626. [30] Larsson, A. L. E.; Persson B. A.; Bäckvall J.-E. Angew. Chem. Int. Ed. Engl.

1997, 36, 1211–1212. [31] a) G. Csjernyik, K. Bogar and J.-E. Bäckvall, Tetrahedron Lett. 2004, 45, 6799–6802. b) Martín-Matute, B.; Edin, M.; Bogár, K.; Bäckvall, J.-E. Angew. Chem. Int. Ed. 2004, 43, 6535–6539. [32] a) Shvo, Y.; Czarkie, D.; Rahamim, Y.; Chodosh, D. F. J. Am. Chem. Soc. 1986, 108, 7400–7402. b) Almeida M. L. S.; Beller M.; Wang, G.-Z.; Bäck- vall, J.-E. Chem. Eur. J. 1996, 2, 1533–1536. c) Samec J. S.; Bäckvall J.-E. Chem. Eur. J. 2002, 8, 2955–2961. d) Éll, A. H.; Samec, J. S. M.; Brasse, C.; Bäckvall, J.-E. Chem. Commun. 2002, 1144–1145. [33] Warner, M. C.; Casey. C. P.; Bäckvall, J.-E. Shvo’s Catalyst in Hydrogen Transfer Reactions. In Bifunctional Molecular Catalysis: Topics in Or- ganometallic Chemistry; Ikariya, T.; Shibasaki, M., Eds.; Springer-Verlag: Berlin, 2011; pp 85–125. [34] Selected examples: a) Martin-Matute, B.; Edin, M.; Bogár, K.; Kaynak, F. B.; Bäckvall, J.-E. J. Am. Chem. Soc. 2005, 127, 8817–8825. b) Träff, A.; Bogár, K.; Warner, M.; Bäckvall, J.-E. Org. Lett. 2008, 10, 4807–4810. c) Borén L.; Leijondahl, K.; Bäckvall, J.-E. Tetrahedron Lett. 2009, 50, 3237–3240. d) Warner, M.; Nagendiran, A.; Bogár, K.; Bäckvall, J.-E. Org. Lett. 2012, 14, 5094–5097. e) Warner, M.; Shevchenko, G. A.; Jouda, S.; Bogár, K.; Bäckvall, J.-E. Chem. Eur. J. 2013, 19, 13859–13864. [35] a) Martín-Matute, B.; Åberg, J. B.; Edin, M.; Bäckvall, J.-E. Chem. Eur. J. 2007, 13, 6063–6072. b) Åberg, J. B.; Nyhlén, J.; Martín-Matute, B.; Privalov, T.; Bäckvall, J.-E. J. Am. Chem. Soc. 2009, 131, 9500–9501. c) Nyhlén, J.; Privalov, T.; Bäckvall, J.-E. Chem. Eur. J. 2009, 15, 5220–5229. d) Warner, M. C.; Verho, O.; Bäckvall, J.-E. J. Am. Chem. Soc. 2011, 133, 2820–2823. [36] Paetzold, J.; Bäckvall, J.-E. J. Am. Chem. Soc. 2005, 127, 17620–17621. [37] Lawrence, S. A. Amines: Synthesis, Properties and Applications; Cambridge University Press: Cambridge, 2006. [38] Ray, P.; Wright, J.; Adam, J.; Boucharens, S.; Black, D.; Brown, A. R.; Epemolu, O.; Fletcher, D.; Huggett, M.; Jones, P.; Laats, S.; Lyons, A.; Rich- ard Morphy, J. M.; Sherborne, B.; Sherry, L.; Straten, N.; Westwood, P.; York, M. Bioorg. Med. Chem. Lett. 2011, 21, 1084–1088. [39] a) Takahata, H; Banba, Y.; Momose, T. Tetrahedron: Asymmetry. 1991, 2, 445–448. b) Pu, L.; Yu, H.-B. Chem. Rev. 2001, 101, 757–824. c) Wee, A. G. H. J. Org. Chem. 2001, 66, 8513–8517. [40] a) Hashimoto, M.; Eda, Y.; Osanai, Y.; Iwai, T.; Aoki, S. Chem. Lett. 1986, 15, 893–896. b) Cossy, J.; Dumas, C.; Gomez Pardo, D. Eur. J. Org. Chem. 1999, 1693–1699. c) Kizaki, N.; Yasohara, Y.; Nagashima, N.; Hasegawa, J. J. Mol. Catal. B: Enzym. 2008, 51, 73–80. d) Lacheretz, R.; Gomez Pardo, D.; Cossy, J. Org. Lett. 2009, 11, 1245–1248. e) Bisogno, F. R.; Cuetos, A.; Orden, A. A.; Kurina-Sanz, M.; Lavandera, I.; Gotor, V. Adv. Synth. Catal.

2010, 352, 1657–1661. [41] a) Hasegawa, J.; Sawa, I.; Mori, N.; Kutsuki, H.; Ohashi, T. Enantiomer 1986, 2, 311–314. b) Horiguchi, A.; Mochida, K. Biosci. Biotech. Biochem. 1995, 59, 1287–1290. c) Tomori, H.; Shibutani, K.; Ogura, K. Bull. Chem. Soc. Jpn. 1996, 69, 207–215. d) Busto, E.; Gotor-Fernández, V.; Gotor, V. Chem. Rev. 2011, 111, 3998–4035. [42] a) Itoh, T.; Han, S.-H.; Matsushita, Y.; Hayase, S. Green Chem. 2004, 6, 437–439. b) Itoh, T.; Matsushita, Y.; Abe, Y.; Han, S.-H.; Wada, S.; Hayase, S.; Kawatsura, M.; Takai, S.; Morimoto, M.; Hirose, Y. Chem. Eur. J. 2006, 12, 9228–9237. c) Han, S.-H.; Hirakawa, T.; Fukuba, T.; Hayase, S.; Ka- watsura, M.; Itoh, T. Tetrahedron: Asymmetry 2007, 18, 2484–2490. d) Abe, Y.; Hirakawa, T.; Nakajima, S.; Okano, N.; Hayase, S.; Kawatsura, M.; Hi- rose, Y.; Itoh, T. Adv. Synth. Catal. 2008, 350, 1954–1958. e) Abe, Y.; Kude, K.; Hayase, S.; Kawatsura, M.; Tsunashima, K.; Itoh, T. J. Mol. Catal. B: Enzym. 2008, 51, 81–85. f) Abe, Y.; Yoshiyama, K.; Yagi, Y.; Hayase, S.; Kawatsura, M.; Itoh, T. Green Chem. 2010, 12, 1976–1980. [43] According to data sheet for Lipase PS “Amano” SD provided by Amano Enzyme Inc. [44] Acetate 14e was obtained in a 1:1 diastereomeric mixture with a high ee. In order to determine the ee more precisely the double bond was reduced, affording the known acetate 14c in an ee of 99%. [45] Rossiter, B. E.; Verhoeven, T. R.; Sharpless, K. B. Tetrahedron Lett. 1979, 20, 47334836. b) Hoveyda, A. H.; Evans, D. A.; Fu, G. C. Chem. Rev. 1993, 93, 1307–1370. c) Adam, W.; Wirth, T. Acc. Chem. Res. 1999, 32, 703–710. d) Grocock, E. L.; Marples, B. A.; Toon, R. C. Tetrahedron, 2000, 56, 989– 992. [46] a) Calaza, M. I.; Hupe, E.; Knochel, P. Org. Lett. 2003, 5, 1059–1061. b) Das, P. P.; Lysenko I. L.; Cha J. K. Angew. Chem. Int. Ed. 2011, 50, 9459– 9461. [47] a) Büchi, G.; Vogel, D. E. J. Org. Chem. 1983, 48, 5408–5409. b) Beaulieu, P.; Ogilvie, W. W. Tetrahedron Lett. 2003, 44, 8883–8885. [48] a) Noyori, R.; Ohkuma, T. Angew. Chem. Int. Ed. 2001, 40, 40–73. b) He, P.; Liu, X.; Zheng, H.; Li, W.; Lin, L.; Feng, X. Org. Lett. 2012, 14, 5134–5137. [49] a) Liu, X.; Li, X.; Chen, Y.; Hu, Y.; Kishi, Y. J. Am. Chem. Soc. 2012, 134, 6136–6139. b) Aikawa, K.; Miyazaki, Y.; Mikami, K. Bull. Chem. Soc. Jpn. 2012, 85, 201–208. [50] Enders, D.; Reichenbach, L. F. Synthesis 2013, 45, 959965. [51] a) Lee, D.; Huh, E. A.; Kim, M.-J.; Jung, H. M.; Koh, J. H.; Park, J. Org. Lett. 2000, 15, 2377–2379. b) Roengpithya, C.; Patterson, D. A.; Gibbins, E. J.; Taylor, P. C.; Livingston, A. G. Ind. Eng. Chem. Res. 2006, 45, 7101– 7109. c) Kim, M.-J.; Lee, H. K.; Park, J. Bull. Korean Chem. Soc. 2007, 28, 2096–2098. d) Bogár, K.; Vidal, P. H.; Alcántara León, A. R.; Bäckvall, J.-E.

Org. Lett. 2007, 9, 3401–3404. e) Norinder, J.; Bogár, K.; Kanupp, L.; Bäckvall, J.-E. Org. Lett. 2007, 9, 5095–5098. [52] Akai, S. Chem. Lett. 2014, 43, 746–754. [53] a) Trost, B. M.; Kulawiec, R. J. J. Am. Chem. Soc. 1993, 115, 2027–2036. b) Ahlsten, N.; Bartoszewicz, A.; Martín-Matute, B. Dalton Trans. 2012, 41, 1660–1670. [54] a) Bäckvall, J.-E.; Andreasson, U. Tetrahedron Lett. 1993, 34, 5459–5462 b) Martín-Matute, B.; Bogár, K.; Edin, M.; Kaynak, F. B.; Bäckvall, J.-E. Chem. Eur. J. 2005, 11, 5832–5842. [55] Bartoszewicz, A.; Livendahl, M.; Martín-Matute, B. Chem. Eur. J. 2008, 14, 10547–10550. [56] Schkeryantz J. M.; Pearson, W. H. Tetrahedron 1996, 52, 3107–3116. [57] See supporting information: Lihammar, R.; Millet, R.; Bäckvall, J.-E J. Org. Chem. 2013, 78, 12114–12120. [58] a) Fuchs, P. L.; Braish, T. F. Chem. Rev. 1986, 86, 903–917. b) Trost, B. M.; Seoane, P.; Mignani, S.; Acemoglu, M. J. Am. Chem. Soc. 1989, 111, 7487– 7500. c) Trost, B. M.; Organ, M. G.; O'Doherty G. A. J. Am. Chem. Soc. 1995, 117, 9662–9670. [59] Fujioka, H.; Matsuda, S.; Horai, M.; Fujii, E.; Morishita, M.; Nishiguchi, N.; Hata, K.; Kita, Y. Chem. Eur. J. 2007, 13, 5238–5248. [60] Johnson, C. R.; Adams, J. P.; Braun, M. P.; Senanayake C. B. W. Tetrahe- dron Lett. 1992, 33, 919–922. [61] Shinokubo, H.; Miki, H.; Yokoo, T.; Oshima, K.; Utimoto, K. Tetrahedron 1995, 51, 11681–11692. [62] a) Hederos, M.; Konradsson, P. J. J. J. Carbohydr. Chem. 2005, 24, 297–320. b) Christina, A. E.; Blas Ferrando, V. M.; de Bordes, F.; Spruit, W. A.; Overkleeft, H. S.; Codée, J. D. C.; van der Marel, G. A. Carbohydr. Res. 2012, 356, 282–287. [63] a) Kochetkov, N. K.; Dmitriev, B. A.; Chizhov, O. S.; Klimov, E. M.; Malysheva, N. N.; Chernyak, A. Y.; Bayramova, N. E.; Torgov, V. I. Carbohydr. Res. 1974, 33, C5–C7. b) H. Mobarak, O. Engström, G. Widmalm, RSC Adv., 2013, 3, 23090–23097. [64] Söderman, P.; Larsson, E. A.; Widmalm, G. Eur. J. Org. Chem. 2002, 1614– 1618. [65] a) Drueckhammer, D. G.; Hennen, W. J.; Pederson, R. L.; Barbas, C. F.; Gautheron, C. M.; Krach, T.; Wong, C. H. Synthesis 1991, 7, 499–525. b) Sakakibara, T.; Ito, T.; Kotaka, C.; Kajihara, Y.; Watanabe, Y.; Fujioka, A. Carbohydr. Res. 2008, 343, 2740–2743. [66] Ramstadius, C.; Träff, A. M.; Krumlinde, P.; Bäckvall, J.-E.; Cumpstey, I. Eur. J. Org. Chem. 2011, 23, 4455–4459. [67] Akai, S.; Tanimoto, K.; Kita Y. Angew. Chem. Int. Ed. 2004, 43, 1407–1410. [68] Åberg, J. B.; Warner, M. C.; Bäckvall, J.-E. J. Am. Chem. Soc. 2009, 131,

13622–13624. [69] In an EXSY experiment, chemical exchange is monitored. A peak is selec- tively excited; after which the magnetization is transferred to the other species involved in the equilibrium during a mixing period. This gives an exchange peak, in addition to the excited peak, in the resulting spectrum. The integral of the exchange peak, and the ratio of the excited and the exchange peak, depends on the length of the mixing period, the exchange rate and the relaxation rate. [70] Borén, L.; Martín-Matute, B.; Xu, Y.; Cordova, A.; Bäckvall J.-E. Chem. Eur. J. 2006, 12, 225–232. [71] The optimized ratio of the enzyme and the two surfactants has been studied previously (ref 70) and showed to be 4:1:1 (enzyme/Brij56/pyranoside). [72] Maiti, D. K.; Halder, S.; Pandit, P.; Chatterjee, N.; Joarder, D. D.; Pramanik, N.; Saima, Y.; Patra, A.; Maiti, P. K. J. Org. Chem. 2009, 74, 8086–8097. [73] Shan, M.; Sharif, E. U.; O’Doherty, G. A. Angew. Chem. Int. Ed. 2010, 49, 9492–9495. [74] a) Mallams, A. K.; Puar, M. S.; Rossman, R. R.; McPhail, A. T.; Gross, P. M.; Macfarlane, R. D.; Stephens, R. L. J. Chem Soc. Perkin Trans. 1 1983, 1497–1534. b) Zhang, H.; White-Phillip, J. A.; Melancon, C. E.; Kwon, H.- J.; Yu, W.-L; Liu, H.-W. J. Am. Chem. Soc. 2007, 129, 14670–14683. [75] a) Brimacornbe, J. S.; Hanna, R.; Saeed, M. A.; Tucker, L. C. N. J. Chem. Soc., Perkin Trans. 1 1982, 2583–2587. b) Braun, M.; Moritz, J. Synlett 1991, 750–752. c) Andreana, P. R.; McLellan, J. S.; Chen, Y.; Wang, P. G. Org. Lett. 2002, 4, 3875–3878. d) Timmons, S. C.; Jakeman, D. L. Carbohydr. Res. 2007, 342, 2695–2704. [76] Tararov, V. I.; Börner, A. Synlett 2005, 203–211. [77] a) Koszelewski, D.; Tauber, K.; Faber, K.; Kroutil W. Trends Biotechnol. 2010, 28, 324–332. b) Mathew, S.; Yun, H. ACS Catal. 2012, 2, 993–1001. [78] a) Denmark, S. E.; Nicaise, O. J. C. Alkylation of Carbonyl and Imino Groups. In Comprehensive Asymmetric Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.; Springer: Berlin, 1999; pp 923–961. b) Mukherjee, S.; Yang, J. W.; Hoffmann, S.; List, B. Chem. Rev. 2007, 107, 5471–5569. c) Nugent, T. C.; El-Shazly, M. Adv. Synth. Catal. 2010, 352, 753–819. [79] Bolm, C.; Hildebrand, J. P.; Muniz, K. Recent Advances in Asymmetric Dihydroxylation and Aminohydroxylation. In Catalytic Asymmetric Synthe- sis; 2 ed.; Ojima, I., Ed.; Wiley-VCH: New York, 2000; pp 399–428. [80] Kim, Y.; Park, J.; Kim, M.-J. Chem. Cat. Chem. 2011, 3, 271. [81] Reetz, M. T., Schimossek, K. Chimia, 1996, 50, 668–669. [82] Parvulescu, A.; De Vos, D.; Jacobs, P. Chem. Commun. 2005, 5307–5309. [83] a) Kim, M.-J.; Kim, W.-H.; Han, K.; Choi, Y. K.; Park, J. Org. Lett. 2007, 9, 1157–1159. b) Cheng, G.; Xia, B.; Wu, Q.; Lin, X. RSC Advances 2013, 3, 9820–9828. c) Jin, Q.; Jia, G.; Zhang, Y.; Li, C. Catal. Sci. Technol. 2014, 4,

464–471. d) Xu, G.; Dai, X.; Fu, S.; Wu, J.; Yang, L. Tetrahedron Lett. 2014, 55, 397–402. [84] Han, K.; Kim, Y.; Park, J; Kim, M.-J. Tetrahedron Lett. 2010, 51, 5581– 5584. [85] Shakeri, M.; Tai, C.-W.; Göthelid, E.; Oscarsson, S.; Bäckvall, J.-E. Chem. Eur. J. 2011, 17, 13269–13273. [86] a) Johnston, E. V.; Verho, O.; Kärkäs, M. D.; Shakeri, M.; Tai, C.-W.; Palmgren, P.; Eriksson, K.; Oscarsson, S.; Bäckvall, J.-E. Chem. Eur. J. 2012, 18, 12202–12206. b) Verho, O.; Nagendiran, A.; Johnston, E. V.; Tai, C.-W.; Bäckvall, J.-E. ChemCatChem. 2013, 5, 612–618. c) Verho, O.; Nagendiran, A.; Tai, C.-W.; Johnston, E. V.; Bäckvall, J.-E. ChemCatChem. 2014, 6, 205–211. [87] Engström, K.; Johnston, E. V.; Verho, O.; Gustafson, K. P. J.; Shakeri, M.; Tai, C.-W.; Bäckvall, J.-E. Angew. Chem. Int. Ed. 2013, 52, 14006–14010. [88] Cammenberg, M.; Hult, K.; Park, S. ChemBioChem, 2006, 7, 1745–1749. [89] Verho, O.; Gustafson, K. P. J.; Nagendiran, A.; Tai, C.-W.; Bäckvall, J.-E. ChemCatChem 2014, 6, 3153–3159. [90] Pellissier, H. in Chirality from Dynamic Kinetic Resolution, The Royal So- ciety of Chemistry, Cambridge, 2011, pp 191–242. [91] Humphrey, C. E.; Ahmed, M.; Ghanem, A.; Turner, N. J. Application of Enzymes in Kinetic Resolutions, Dynamic Kinetic Resolutions and Deracemization Reactions. In Separation Of Enantiomers: Synthetic Meth- ods; M. Todd, Ed.; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, 2014. [92] Sorbera, L. A.; Castañer, R.M.; Castañer, J. Drugs Future 2000, 25, 907–916. [93] a) Bymaster, F. P.; Beedle, E. E.; Findlay, J.; Gallagher, P. T.; Krushinski, J. H.; Mitchell, S.; Robertson, D. W.; Thompson, D. C.; Wallace, L.; Wong, D. T. Bioorg. Med. Chem. Lett. 2003, 13, 4477–4480. b) Hunziker, M. E.; Suehs, B. T.; Bettinger, T. L.; Crismon, M. L. Clinical Theraputics 2005, 27, 1126–1143. c) Mariappan, P.; Alhasso, A.; Ballantyne, Z.; Grant, A.; N’Dow, J. Eur. Urol. 2007, 51, 67–74. c) Cashman, J. R.; Ghirmani, S. Bioorg. Med. Chem. 2009, 17, 6890–6897. d) Norman, T. R.; Olver, J. S. Drug Design, Development and Therapy. 2010, 4, 19–31. [94] http://www.pmlive.com/top_pharma_list/Top_50_pharmaceutical_ products_by_global_sales (access Oct 2014) [95] a) Deeter, J.; Franzier, J.; Staten, G.; Staszak, M.; Wiegel, L. Tetrahedron Lett. 1990, 31, 7101–7104. b) Wheeler, W. J.; Kuo, F. J. Labelled Compd. Radiopharm. 1995, 36, 213–223. c) Genov, D. G.; Ager, D. J. Angew. Chem. Int. Ed. 2004, 43, 2816–2819. d) Wang, G.; Liu, X.; Zhao, G. Tetrahedron: Asymmetry 2005, 16, 1873–1879. e) He, S. Z.; Li, X. M.; Dai, J.; Yan, M. Chin. Chem. Lett. 2008, 19, 23–25. f) Buitrago, E.; Lundberg, H.; Andersson, H.; Ryberg, P.; Adolfsson, H. ChemCatChem 2012, 4, 2082–2089.

[96] a) Liu, H.; Hoff, B. H.; Anthonsen, T. Chirality 2000, 12, 26–29. b) Sakai, K.; Sakurai, R.; Yuzawa, A.; Kobayashi, Y.; Saigo, K. Tetrahedron: Asym- metry 2003, 14, 1631–1636. c) Kamal, A.; Malik, M. S.; Shaik, A. A.; Azeeza, S. Tetrahedron: Asymmetry 2008, 19, 1078–1083. d) Kamal, A.; Malik, M. S.; Shaik, A. A.; Azeeza, S. J. Mol. Cat. B: Enzym. 2009, 58, 132– 137. [97] Fujima, Y.; Ikunaka, M.; Inoue, T.; Matsumoto, J. Org. Process Res. Dev. 2006, 10, 905–913. [98] a) Pàmies, O.; Bäckvall, J.-E. Adv. Synth. Catal. 2001, 343, 726–731. b) Pàmies, O.; Bäckvall, J.-E. Adv. Synth. Catal. 2002, 344, 947–952. [99] a) Kamal, A.; Khanna, G. B. R.; Ramu, R.; Krishnaji, T. Tetrahedron Lett. 2003, 44, 4783–4787. b) Turcu, M. C.; Perkiö, P.; Kanerva, L. T. ARKIVOC, 2009, 3, 251–263. [100] This change in selectivity in the transesterification is due to a single mutation of the enzyme. By replacing a sterically demanding tryptophan (W) to the smaller amino acid alanin (A) an increase of space in the active site was achieved. This extra space makes it possible for the enzyme to accommodate the substrate in a new fashion, which inverted its selectivity. However, a longer side chain is usually needed in order for the enzyme to display high selectivity. Clearly the methylene-nitrile moiety was not large enough. [101] a) Magnusson, A. O.; Rotticci-Mulder, J. C.; Santagostino, A.; Hult, K. ChemBioChem 2005, 6, 1051–1056. b) Magnusson, A. O.; Takwa, M.; Hamberg, A.; Hult, K. Angew. Chem. Int. Ed. 2005, 44, 4582–4585. c) Engström, K.; Vallin, M.; Syren, P.-O.; Hult, K.; Bäckvall, J.-E. Org. Biomol. Chem. 2011, 9, 81–82. [102] Verho, O.; Johnston, E. V.; Karlsson, E.; Bäckvall, J.-E. Chem. Eur. J. 2011, 17, 11216–11222. [103] Brown, H. C.; Cohi, Y. M.; Narasimhan, S. J. Org. Chem., 1982, 47, 3153– 3163. [104] a) Anderson, E. M.; Larsson, K. M.; Kirk. O. Biocatal. Biotransformation 1998, 16, 181–204. b) Kirk, O.; Würtz Christensen, M. Org. Process Res. Dev. 2002, 6, 446–451. [105] a) Uppenberg, J.; Patkar, S.; Bergfors, T.; Jones, T. A. J. Mol. Biol. 1994, 235, 790–792. b) Uppenberg, J.; Öehrner, N.; Norin, M.; Hult, K.; Kleywegt, G. J.; Patkar, S.; Waagen, V.; Anthonsen, T.; Jones, T. A. Biochem. 1995, 34, 16838–16851. [106] Rotticci, D.; Hæffner, F.; Orrenius, C.; Norin, T.; Hult K. J. Mol. Catal. B, 1998, 5, 267–272. [107] The starting substrate is a racemic mixture of diastereomers and the E value is only applicable on enantiomers. However, it was found that the stereochemistry of the sterically hindered alcohol did not affect the enzyme- catalyzed acetylation of the less hindered alcohol. The measured kinetic pa-

rameter was called the “pseudo E value”. [108] Martín-Matute, B.; Edin, M.; Bäckvall, J.-E. Chem. Eur. J. 2006, 12, 6053– 6061. [109] Nyhlen, J.; Martin-Matute, B.; Sandström, A. G.; Bocola, M.; Bäckvall, J.-E. ChemBioChem 2008, 9, 1968–1974. [110] a) Edin, M.; Bäckvall, J.-E. J. Org. Chem. 2003, 68, 2216–2222. b) Ema, T.; Ura, N.; Yoshii, M.; Korenaga, T.; Sakai, T. Tetrahedron, 2009, 65, 9583– 9591. [111] Schmitke, J. L.; Wescott, C. R.; Klibanov, A. M. J. Am Chem. Soc. 1996, 118, 3360–3365 [112] Wangikar, P. P.; Graycar, T, P.; Estell, D. S.; Dordick, J. S. J. Am. Chem. Soc. 1993, 115, 12231–12237 [113] Garber, M.; Bousquet-Dubouch, M. P.; Lamare, S.; Legoy, M.-D. Biochim. Biophys. Acta – Proteiins Proteomics 2003, 1648, 24–32 [114] Léonard, V.; Fransson, L.; Lamare, S.; Hult. K.; Graber, M. ChemBioChem 2007, 8, 662–667.