Enantioselective Alcohol
Synthesis using
Ketoreductases, Lipases or
an Aldolase
Cover: Representing “Diversity in Parameter Space”
Enantioselective Alcohol
Synthesis using
Ketoreductases, Lipases or
an Aldolase
Proefschrift
ter verkrijging van de graad van doctor aan de Technische Universiteit Delft,
op gezag van Rector Magnificus, prof. dr. ir. J.T. Fokkema, voorzitter van het College van Promoties,
in het openbaar te verdedigen op 29 mei 2006 om 12.30 uur
door
Menno Jort SORGEDRAGER ingenieur in de bioprocestechnologie
Dit proefschrift is goedgekeurd door de promotor: Prof. dr. R.A. Sheldon
Toegevoegd promotor: Dr. ir. F. van Rantwijk
Samenstelling promotiecommissie:
Rector Magnificus Voorzitter
Prof. dr. R.A. Sheldon Technische Universiteit Delft, promotor
Dr. ir. F. van Rantwijk Technische Universiteit Delft, toegevoegd promotor Prof. dr. W.R. Hagen Technische Universiteit Delft
Prof. dr. J.A.M. de Bont Technische Universiteit Delft Prof. dr. A. Liese Technische Universiteit Hamburg Prof. dr. ir. A.P.G. Kieboom Universiteit van Leiden
Dr. G. Huisman Codexis Inc. (USA, CA)
The research described in this thesis was financially supported and performed in cooperation with Codexis inc. (Redwood City, USA).
ISBN: 90-9020702-3
Copyright 2005 by M.J. Sorgedrager
INTRODUCTION
Chapter 1 Introduction 1
PART I: LIPASE CATALYSED RESOLUTION OF ALDOL ADDUCTS Chapter 2 Lipase catalysed Resolution of Nitro aldol Adducts 21 Chapter 3 Optimising the Deracemisation of Nitro aldol Adducts 33 PART II ENANTIOSELECTIVE CARBONYL REDUCTION Chapter 4 Asymmetric Reduction with Candida Magnoliae Ketoreductase S1 47 Chapter 5 Asymmetric Carbonyl-Reductions with Microbial Ketoreductases 63 PART III ENANTIOSELECTIVE ALDOL REACTION: DERA Chapter 6 Production and Optimisation of 2-deoxyribose-5-phosphate Aldolase 79 Chapter 7 DERA as Catalyst for Statin Precursors 97 Chapter 8 Cross-linked Enzyme Aggregates of DERA 113 Summary 125
Samenvatting 127
Dankwoord 129
1
Introduction
2
Introduction
The carbonyl group is probably the most important functional group in organic chemistry. Besides aldehydes and ketones, many functional groups contain the C=O bond, to which they own much of their behaviour in chemistry, such as carboxylic acids, acidhalides, acid anhydrides, esters and amides. When the molecule is bearing at least one α-hydrogen, which is acidic due to the electron-withdrawing effect of the carbonyl group and therefore can be abstracted by a strong base, these compounds are able to react via the resulting enolate ion. This forms the basis for a variety of synthetically usefull C-C bond forming reactions. Michael addition, aldol and Claissen reactions are examples of this reaction pathway.
One of the most important of these reactions is probably the aldol reaction. Besides taking a central role in synthetic organic chemistry, this reaction is vital for all living organisms in nature. Aldol-type reactions are the key step in the metabolic pathways of micro-organisms, where it takes part in the breakdown of carbohydrates. Besides its role in anabolism, the aldol reaction is often applied in catabolism as well. The aldol reaction is initiated by the formation of the enolate by a base, followed by nucleophilic attack hereof on another carbonyl carbon (Figure 1.1). Depending on the reaction conditions it is possible to isolate the β-Hydroxy carbonyl compound or the unsaturated product that is formed by subsequent dehydration. In the latter case it is referred to as aldol condensation.
β-Hydroxy carbonyl compounds are versatile and interesting intermediates. They can easily be synthesised directly via an aldol reaction of the corresponding aldehydes. The β-ketoesters, later described in this Chapter in connection with their reduction into β-Hydroxy esters, are easily accessible via the aldol reaction analogue for esters known as the Claisen condensation.
O O O H2C O --H 2C O H H Base OH O dehyration
The demand for optically pure intermediates has grown rapidly in recent years. Therefore it is of growing importance to control chirality during a reaction. Although there are various chemical methods available1 biocatalysts are more and more applied to achieve this goal.1,2 Enzymes from the classes hydrolases, oxidoreductases and lyases are synthetically interesting for this purpose. Lipases, able to hydrolyze esterbonds; Ketoreductases, able to reduce carbonyl groups and aldolases that perform aldol reactions are well suited to synthesize enantiopure secondary alcohols.
Biocatalysis: Kinetic resolution vs asymmetric synthesis
Two strategies are possible to obtain homochiral compounds. One is a kinetic resolution of a racemic mixture, in which the enzyme catalyses the reaction of the two enantiomers of the substrate with different rates. The lipase mediated esterification of racemic secondary alcohols (Chapters 2 and 3) is a well known example of applying a kinetic resolution strategy. The other is an asymmetric synthesis starting from a prochiral substrate. In an asymmetric synthesis a new chiral centre is introduced into the substrate molecule. Examples of asymmetric syntheses, which are reported in this thesis, are the asymmetric reduction mediated by ketoreductases (Chapters 4 and 5) and the DERA catalysed aldol reaction (Chapters 6-8).
The efficiency of a resolution process is dependent on the difference in rate in which the two enantiomers are converted. This is indicated with the enantiomeric ratio (E), which is the ratio of the two pseudo-first order kinetic rate constants (eq. 1.1). The enantiomeric excess of a compound (ee) is the excess amount of one enantiomer compared to the total amount of both enantiomers (eq. 1.2).
The enantiomeric ratio:
Introduction
4
The enantiomeric excess:
(
)
(
RR SS)
c c ee c c − = + (eq. 1.2) cx: Concentration enantiomer xThe E value can be determined from experimental data via the conversion and either the ee of the substrate or the ee of the product.3
Via substrate:
{
(
)(
)
}
(
)(
)
{
}
ln 1 1 ln 1 1 s s ee E ee ξ ξ − − = − + (eq. 1.3) Via product:{
(
)
}
(
)
{
}
ln 1 1 ln 1 1 p p ee E ee ξ ξ − + = − − (eq. 1.4) E: Enantiomeric ratioees: Enantiomeric excess of the substrate eep: Enantiomeric excess of the product
ξ: Conversion
An alternative method for calculating the enantiomeric ratio (E) directly from the enantiomeric excess of the substrate (ees) and the product (eep) is given
by equation 1.5. The conversion of an irreversible reaction without substrate inhibition (up to 40% conversion) can then be calculated directly from the optical purities of the substrate (ees) and the product (eep) according to
equation 1.6.3
Direct from ee’s:
(
)
(
1)
(
(
1)
)
ln ln 1 1 s s s p s p ee ee E ee ee ee ee − + = + + (eq. 1.5) Conversion: s(
)
s p ee ee ee ξ = + (eq. 1.6)(DKR, Figure 1.2) The biggest challenge here is to find conditions that are compatible with both processes.
(S)-Substrate (S)-Product
(R)-Substrate Slow (R)-Product
Fast
in Situ Racemisation
(S)-Substrate (S)-Product
(R)-Substrate Slow (R)-Product
Fast
Classical kinetic resolution Dynamic kinetic resolution
Figure 1.2 General schemes for a classical kinetic resolution and a dynamic kinetic resolution
Lipase resolution
Lipases are enzymes of the hydrolase group. In nature they catalyse the hydrolysis of triglycerides and other long chain fats and oils. Besides their normal substrates most lipases accept a very broad range of acyl donors and acyl acceptors5. Since lipases generally exhibit mostly a high stability, accept a wide variety of substrates and have a wide occurrence in nature, they have become readily available and industrially applicable enzymes. They have current industrial applications as detergent enzymes, in food and paper technology and in the pharmaceutical and speciality chemicals industry5.
Introduction 6 Asp O O N N His Ser O NH O O HN Asp O O N N His Ser O NH O O HN Ser O NH O O HN ROAc O Acyl enzyme intermediate (with alcohol) Tetrahedral intermediates Free enzyme (with acyldonor) H H O -H H O Asp O O N N His H O O O R O H Asp O O N N His Ser O NH O O HN O -H H O R
Figure 1.3 Catalytic machanism of a lipase catalyzed esterification of a chiral alcohol with
vinyl acetate as model acyl donor.
the next step this reacts with a nucleophile (in Figure 1.3 an alcohol) to result in the product that is released and the enzyme that is ready for the next catalytic cycle5,6,7.
Often the enantioselective catalytic properties of lipases are used in a kinetic resolution to resolve chiral substrates such as alcohols or amines. Recently these processes are more often designed as a dynamic kinetic resolution (see previous paragraph).
Keto reductases
The asymmetric reduction of ketones is one of the most efficient methods for the production of chiral secondary alcohols. In nature such reactions are important steps in many metabolic pathways and therefore the source of carbonyl reducing enzymes from nature is abundant. Keto reductases belong to the enzyme class of the oxidoreductases. The aldo-keto reductase (AKR) superfamily and the short chain dehydrogenase/reductase (SDR) superfamily, to which keto reductases belong, are subdivisions of this class of enzymes. Based on crystal structure analysis, sequence information and point mutation studies a common catalytic pathway for SDR enzymes has been proposed. The majority of SDR enzymes have an active site that consists of a catalytic triad of tyrosine, lysine and serine (Figure 1.4). 8,9
Introduction
8
protonation of the catalytic tyrosine residue by bulk water to bring the enzyme back in its apo state. (Figure 1.4).
N O OH H H OH R H2N O NAD(P)H Lys +H 3N Tyr HO Ser OH N O OH H H OH R H2N O NAD(P)H R O Lys +H 3N Tyr HO Ser OH N O OH OH R H2N O NAD(P)+ R O Lys +H 3N Tyr O Ser OH H H H N O OH OH R H2N O NAD(P)+ Lys +H 3N Tyr O Ser OH H HOH HOH Lys +H 3N Tyr HO Ser OH NAD(P)+ NAD(P)H E: Apo state
ER: Binary reduced state ERK: Ternary keto-bound reduced state
EOA: Ternary alcohol-bound oxidized state EO: Binary oxidized state
OH -R O R OH H
Figure 1.4: Proposed catalytic mechanism for SDR enzymes
Enzymes of the SDR superfamily use four different pathways for transferring the hydride from the cofactor NAD(P)H to the carbonyl carbon of the substrate. Either the pro-(S) (for E2 and E4 enzymes) or the pro-(R) hydride from the cofactor (for E1 and E3 enzymes) can be transferred to either the
N R HR HS H2N O NAD(P)H N R HR HS H2N O NAD(P)H s L O Re face Si face E1 E2 E4 E3
Figure 1.5 Stereospecificity of SDR enzymes
Most of the ketoreductases have a preference for one of the two possible cofactors: NADH, NADPH. The origin of SDRs having different coenzyme specificity is related to their natural function. The intracellular ratio of NADPH over NADP+ is relatively high. Hence NADP(H) dependent enzymes function
in vitro mainly as reductases while, due to a significantly low NADH over
NAD+ ratio enzymes dependent on NAD(H) will mainly work as dehydrogenases.12
The specificity of the SDR enzymes towards NAD(H) or NADP(H) as coenzyme seems to be dominated by electrostatic effects.13,14,15,16,17 The amino acids located around the binding position of the adenine ribose moiety of the coenzyme are typically different in nature for NAD(H) preferring enzymes than for NADP(H) preferring enzymes.
Introduction 10 OH 2-O 3PO O 2-O 3PO R O OH OH OH OH 2-O 3PO N Lys OH 2-O 3PO NH Lys OH 2-O 3PO NH Lys O 2-O 3PO OH Type I Type II OH OPO3 2-O OH OH HO O H2N Zn His His His H Glu CO2 OH OPO3 2-O OH OH HO O H2N Zn His His His Glu CO2H OH OPO3 2-O OH OH HO O H2N Zn His His His OH O R=H R=OPO3 2-Glu CO2H Asp CO2H Asp CO2H
Figure 1.6 Mechanistic scheme of the two types of aldolases represented by
fructose-1,6-bisphosphate aldolase (Type I) and fuculose-1-phosphate aldolase (type II).
Most aldolases are very specific for and limited by their natural substrates. Further characterization of aldolases is therefore done by classification with the donor substrate they are dependent on (Table 1.1).18 This specificity limits
the range of applications but is beneficial in cross-aldol reactions. Chemically catalyzed cross-aldol reactions mostly yield mixtures comprised of all possible product combinations of the substrate aldehydes and or ketones. Aldolases, owing to their often high selectivity in accepted donor substrates, generally yield a single product.
Table 1.1 Aldolase classification by donor dependence18
Classification Donor substrates Example aldolase Pyruvate dependent pyruvate NeuAc aldolase Phosphoenolpyruvate
dependent
phosphoenolpyruvate NeuAc syntetase
Due to the position of most aldolases in metabolism the donor or the acceptor substrate is a phosphorylated compound. The cell has phosphorylated these molecules somewhere in the preceding metabolic pathway, to make them more reactive, but more important that they cannot leak out of the cell through the cell membrane after prior energy investment. For organic synthetic purposes the necessity for these phosphorylated substrates is a drawback because they are costly to synthesize and after the reaction the phosphate group has to be removed. This has precluded a widespread application of aldolases untill today.
2-Deoxy-D-ribose-5-phosphate aldolase (DERA) is a type I aldolase, which
takes a crucial role in the anabolism of a cell. DERA’s natural reaction is the aldol reaction between acetaldehyde and D-glyceraldehyde-3-phosphate to
form 2-deoxy-D-ribose-5-phosphate, which is the sugar moiety of the DNA
backbone. (Figure1.7) CH3 O OPO3 2-OH O + OPO3 2-OH OH O O OH OH O3PO DERA
2-Figure 1.7 Natural reaction of DERA: the aldol reaction of acetaldehyde and D
-glyceraldehyde-3-phosphate to 2-deoxy-D-ribose-5-phosphate
The crystal structure and the full catalytic mechanism of this aldolase have recently been elucidated19,20 (Figure 1.8) Two lysine residues and an aspartate are identified to be crucial for catalytic activity. Due to electrostatic interactions with the environment of Lys167 has a slightly lower pKa.
Introduction 12 NH3 Lys 201 NH2 Lys 167 O O Asp 102 O Me HOH NH2 Lys 201 HN Lys 167 OH O Asp 102 H Me HOH HO NH2 Lys 201 N Lys 167 OH O Asp 102 H C H2 H OH H2O NH3 Lys 201 HN Lys 167 O O Asp 102 H CH2 HOH O R H2O NH3 Lys 201 N Lys 167 O O Asp 102 H HOH OH R NH3 Lys 201 HN Lys 167 O O Asp 102 H HOH OH HO R O Me R OH O HN Lys 167 H Me HOH HO O O Asp 102 NH3 Lys 201 H O R
Figure 1.8 Catalytic machanism of DERA
While other aldolases only accept certain ketones as aldol donor, DERA is the only known aldolase that accepts aldehydes and ketones as donor substrates. The accepted aldol donors are: acetaldehyde, propanal, acetone and fluoroacetone21. A wide variety of aldol acceptors are possible, albeit with relatively low reaction rates for non-natural acceptors21. With these characteristics DERA has a wide and interesting product scope, like β-hydroxy aldehydes, deoxy sugars, azido sugars or trideoxyhexoses22,23. Recent reviews give a good overview about the possibilities.18,24
Precursors for statins
cholesterol biosynthetic pathway, 3-hydroxy-3-methyl-glutaryl-CoA (HMG-CoA) reductase, was a good target for investigation.25
Compactin and lovastatin, natural products with a powerful inhibitory effect on HMG-CoA reductase, were discovered in the 1970s. These substances were taken into clinical development as potential drugs for lowering LDL cholesterol. The use of compactin was discarded along the way due to serious side effects but large-scale trials confirmed the effectiveness of lovastatin and the observed tolerability of this statin continued to be excellent. Lovastatin was approved by the US FDA in 1987. Several other HMG-CoA reductase inhibitors, now widely known as statins, subsequently became available for prescription.25 Statin medicines that are now on the pharmacy shelves in the U.S. are shown in Figure 1.9.
O H O O HO O Lovastatin (Mevacor®) O H O O HO O Simvastatin (Zocor®) O HO H O O HO O Pravastatin (Pravachol®) OH OH ONa O N F Fluvastatin (Lescol®) N OH OH O -O F O NH 2 Ca2+ Atorvastatin (Lipitor®) OH OH OH O N N N S O O F Rosuvastatin (Crestor®) OH OH O–) 2Ca++ O N F Pitavastatin (Lipalo®)
Figure 1.9 Statins with 3,5-dihydroxy hexanoic acid based side chain26,27
Introduction
14
billion (2002: US$ 8.0 billion) and US$ 6.1 billion (2002: US$ 6.2 billion) are recorded for atorvastatin and simvastatin, respectively. Simvastatin, pravastatin, and lovastatin are produced by fermentation or partial synthesis. Atorvastatin, fluvastatin and rosuvastatin are fully synthetic statins that are rapidly increasing in market value. Pitavastatin is the next synthetic statin, which is in phase 3 of its clinical trials.27
The statins have in common that they have the same homochiral side chain which has the form of a 3,5-dihydroxy acid. Variations are in the hetero aromatic or cyclic residue.27 Since the synthetic statins have such an extremely high market value and require high chemical and stereochemical purity (> 99.5% ee, > 99% de) competition between different research and development groups to find methods to produce the homochiral 3,5-dihydroxy acid side chain precursor, is fierce.27 This makes that these molecules are the
most important homochiral β-hydroxy carbonyl compounds regarding annual sales. Hence various chemo-enzymatic routes for their synthesis utilising the three major classes of synthetically useful enzymes (e.g. hydrolases, oxidoreductases and lyases) have been developed (Figure 1.10).
O OH OEt X O OH OH OR OR' O OH O OEt HO O OH OH NC O OH OEt NC X OH OH O O O OEt X O O OH OtBu HO O O O OEt R'O O OR'' O OEt EtO OH CN NC O O OEt Cl O X H O CH3 H
+
2 1. ADH 2. Lipase Lipase Nitrilase 1. KRED 2. HHDH Aldolase ADH ADHFigure 1.10 Biocatalytic transformations in the synthesis route towards (3R,
Directed evolution
Enzymes in general catalyse their native reaction with very high specificity and enatioselectivity. As a result of natural evolution they are optimized to meet the cell’s demands to perform a specific biological role efficiently. High levels of sophistication have been established over several millions of years of evolution for this specific role. However, the enzyme’s activity, selectivity and stability often do not meet the synthetic chemist’s needs in non-natural reactions and environments. Therefore, there is a great need to adapt and alter the enzyme to meet these demands. Several rational methods are developed, such as site directed mutagenesis28, which targets a specific amino acid residue and exchanges it with another and saturation mutagenesis, where one target amino acid is exchanged by all other natural amino acids. These methods need a high level of knowledge about structure, mechanism, function, etc.
Following the principle of natural evolution, that generates large numbers of variants and selects the “fittest” candidate, a lab scale mimic, called directed evolution, has been developed29,30. Directed evolution is a powerful method to adapt enzymes to the needs of the synthetic chemist without the requirement of detailed knowledge about its crystal structure or the catalytic mechanism.31,32 Directed evolution is an iterative process in which generation of diversity is followed by screening of the resulting library for a desired function and using the positive hits in the next cycle (Figure 1.11)
Introduction
16
Figure 1.11 General scheme for directed evolution using non-recombinative methods (left) or recombinative methods (right)
Scope of this thesis
The work described in this thesis deals with biocatalytic routes to produce homochiral β-hydroxy carbonyl compounds. This type of building block is important in the production of a wide variety of fine chemicals. Three different biocatalytic routes were investigated: a lipase catalysed resolution of the secondary alcohols, the ketoreductase catalysed reduction of the corresponding ketones and a DERA catalysed aldol reaction. Directed evolution aimed at altering the enzyme’s specificity and improving its characteristics to meet the organic synthetic needs takes a central role in the routes that utilize ketoreductases and the aldolase DERA. Analysis of enzyme structure relationships for improved variants produced by directed evolution
wild type gene
Generation of Diversity Expression of variants Selection Non-recombinative method - ePCR
- site saturation mutagenesis - casete mutagenesis Recombinative method - DNA-shuffling™ - StEP - random-priming recombination -hetroduples recombination -ITCHY
Repeat cycle starting with improved variant gene
wild type gene
Generation of Diversity Expression of variants Selection Non-recombinative method - ePCR
- site saturation mutagenesis - casete mutagenesis Recombinative method - DNA-shuffling™ - StEP - random-priming recombination -hetroduples recombination -ITCHY
can provide a basis for understanding mechanistic details and for further improvemend in substrate specificity, activity, selectivity, etc.
References
1 Sheldon, R.A. Chirotechnology: Industrial synthesis of optically active compounds,
1993, Marcel Dekker Inc., New York
2 Panke, S.; Held, M.; Wubbolts, M. Curr. Opin. Biotechnol. 2004, 15, 272-279 3 Straathof, A.J.J.; Jongejan, J.A. Enzym. Microb. Technol. 1997, 21, 559-571 4 Pellissier, H. Tetrahedron 2003, 59, 8291-8327
5 Schmid, R.D.; Verger, R. Angew. Chem. Int. Ed. 1998, 37, 1608-1633 6 Ema, T. Curr. Org. Chem. 2004, 8, 1009-1025
7 Pleiss, J.; Fisher, M.; Schmid, R.D. Chem. Phys. Lipids 1998, 93, 67-80
8 Varughesem K.I.; Xuong, N.H.; Kiefer, P.M.; Matthews, D.A.; Whiteley, J.M. Proc. Natl.
Acad. Sci 1994, 91, 5582-5586
9 Jörnvall, H.; Persson, B.; Krook, M.; Penning, T.M. Biochemistry 1995, 34, 6003-6013 10 Bradshaw, C.W.; Fu, H.; Shen, G.J.; Wong, C.H. J. Org. Chem. 1992, 57, 1526-1532 11 Prelog, V. Pure Appl. Chem. 1964, 9, 119
12 Tanaka, N.; Nonaka, T.; Nakamura, K.T.; Hara, A. Curr. Org. Chem. 2001, 5, 89-111 13 Tanaka, N.; Nonaka, T.; Nakanishi, M.; Deyashiki, Y.; Hara, A.; Mitsui, Y. Structure
1996, 4, 33-45
14 Tsigelny, I.; Baker, M.E. Biochem. Biophys. Res. Commun. 1996, 226, 118-127
15 Nakanishi, M.; Kakumoto, M.; Matsuura, K.; Deyashiki, Y.; Tanaka, N.; Nonaka, T.; Mitsui, Y.; Hara, A. J. Biochem. 1996, 120, 527-263
16 Nakanishi, M.; Matsuura, K.; kaibe, H.; Tanaka, N.; Nonaka, T.; Mitsui, Y.; Hara, A. J.
Biol. Chem. 1997, 272, 2218-2222
17 Mazza, C.; Breton, R.; Housset, D.; Fontecilla-Camps, J.C. J. Biol. Chem. 1998, 273, 8145-8152
18 Machajewski, T.D. Angew. Chem. Int. Ed. 2000, 39, 1352-1374
19 Heine, A.; DeSantis, G.; Luz, J.G.; Mitchell, M.; Wong, C.H. Wilson, I.A. Science 2001, 294, 369-374
20 Liu, J.J.; DeSantis, G.; Wong, C.H. Can. J. Chem./Rev. Can. Chim. 2002, 80, 643-645 21 Barbas, C.F.; Wang, Y.F.; Wong, C.H. J. Am. Chem. Soc., 1990, 112, 2013-2014 22 Gijsen, H.J.M.; Wong, C.H. J. Am. Chem. Soc. 1995, 117, 7585-7591
Introduction
18
25 Tolbert, J.A. Nat. Rev. Drug Discov. 2003, 2, 517-526
26 Öhrlein, R.; Baisch, G. Adv. Synth. Catal. 2003, 345, 713-715 27 Müller, M. Angew. Chem. Int. Ed. 2005, 44, 362-365
28 Trower, M.K. In vitro mutagenesis protocols 1996, Humana Press., New jersey 29 Stemmer, W.P.C. Nature 1994, 370, 389-391
30 Arnold, F.H. Chem. Eng. Sci. 1996, 51, 5091-5102
31 Jaeger, K.; Reetz, M.T. Curr. Opin. Chem. Biol. 2000, 4, 68-73
32 Petrounia, I.P.; Arnold, F.H. Curr. Opin. Biotechnol. 2000, 11, 325-330
33 Jaeger, K.E.; Eggert, T.; Eipper, A.; Reetz, M.T. Appl. Microbiol. Biotechnol. 2001, 55, 519-530
34 Stemmer, W.P.C. Proc. Natl. Acad. Sci.1994, 91, 10747-10751
35 Zhao, H.; Giver, L.; Shao, H.; Affholter, J.A.; Arnold, F.H. Nat. Biotechnol. 1998, 16, 258-261
Part I
2
Lipase catalysed Resolution
of Nitro aldol Adducts
Abstract
The kinetic resolution of a range of 1-nitro-2-alkanols by lipase-catalysed esterification using various lipases and succinic anhydride as acyl donor was studied. E values up to 100 were obtained with Novozym 435 in the resolution of 1-nitro-2-pentanol with succinic anhydride in TBME. Acylation with succinic anhydride was much more enantioselective than with vinyl acetate.
The content of this chapter has been published in:
Lipase catalysed Resolution of Nitro-aldol Adducts
22
Introduction
The aldol reaction is one of the most important methods for C-C bond formation.1 In the related nitro-aldol reaction, the Henry reaction, the coupling of a nitro alkane to an aldehyde or ketone results in the corresponding β-nitro alcohol. Such chiral nitro alcohols are interesting building blocks in organic synthesis. They can be converted into the chiral β-hydroxy amine by reduction of the nitro group. Alternatively, further carbon-carbon bond formation on the α-carbon of the nitro group will lead to a wide variety of other useful intermediates.2,3 Control of the stereochemistry is of importance for many
synthetic purposes, especially for pharmaceutical and agricultural applications.
The products of Henry reactions are secondary alcohols. Hence, they are suitable substrates for resolution by lipase-catalysed enantioselective acylation (Figure 2.1). The lipase-mediated kinetic resolution of secondary alcohols has been widely studied over the past 20 years4 and has become a common synthetic and industrial methodology for producing chiral compounds as pure enantiomers.5,6 + H3C NO2 NO2 OH Base 1c 2c O NO2 OH Lipase Acyl donor NO2 OH NO2 O + O R 2c (S)-3c R = CH3 C2H4COOH (R)-2c
Figure 2.1 Nitro-aldol reaction of 1-nitro-2-pentanol followed by lipase-catalysed resolution
The resolution of some β-nitro alcohols via transesterification with vinyl acetate as the acyl donor has been reported previously,3,7,8 but the separation of the enantiomerically enriched ester and alcohol is often laborious. Acylation with a cyclic anhydride, which also reacts irreversibly, results in a half ester that can be readily extracted from the reaction mixture. This potential benefit for reaction work-up procedures has been reported for the resolution of several secondary alcohols.9,10,11,12,13,14
We have studied the applicability of succinic anhydride as acyl donor in the lipase-mediated resolution of a number of alkyl- and phenylalkyl substituted nitro alcohols. The effects of the lipase and the reaction medium will be discussed.
Results and discussion
Resolution of nitro alcohols: lipases, acyl donors, solvents
The acylation of 1-nitro-2-pentanol (2c, Figure 2.1), which we selected as a suitable test reactant, was performed in the presence of a range of microbial lipases (Table 2.1). The reaction was fast and enantioselective when performed in the presence of Novozym 435, an immobilized preparation of
Candida antarctica lipase B (CaLB). Two cross-linked preparations of CaLB,
the cross-linked enzyme aggregate (CLEA) and the cross-linked enzyme crystal (ChiroCLEC CaB) were much less active, although comparable amounts of units were used,∗ and the enantioselectivity was low. In these hydrophilic particles, with a high density of active protein, the reaction is probably no longer kinetically controlled but limited by the rate of diffusion in and out of the particles. This could lead to lower conversion and enantioselectivity.
Lipase catalysed Resolution of Nitro-aldol Adducts
24
Table 2.1 The performance of different lipases in the resolution of 1-nitro-2-pentanola Enzyme Conversion (%, 24 h) ees (%) eep (%) E Novozym 435 54 92 93 82 (S) CaLB CLEC 11 8 67 5 (S) CaLB CLEA 7 0 5 1 (S) CaLA SP 526 10 11 82 12 (R) BcL (Amano 1) 45 34 46 4 (R) BcL CLEC 54 59 37 4 (R) CrL 3 2 14 1 (R) CrL CLEC 8 1 3 1 (S) ClL nra - - - RmL SP 524 nr - - -
CaLB: Candida anctarctica lipase B; CaLA: Candida antarctica lipase A; RmL: Rhizomucor miehei lipase; CrL: Candida rugosa lipase; ClL: Candida lipolytica lipase, BcL: Burkholderia cepacia lipase Reaction conditions: 1 ml diisopropyl ether, 100 mM 1-nitro-2-pentanol, 100 mM succinic anhydride, 25 mg enzyme or 5 mg CLEC or CLEA.
a No reaction
In the presence of CaLB, the (S)-enantiomer was preferentially converted, as predicted by the Kazlauskas rule.15 Two preparations of Burkholderia cepacia lipase (Amano 1 and ChiroCLEC PC), in contrast, showed a slight preference for the (R)-enantiomer of the reactant. The activity of the other lipases tested was much lower and, apart from Candida antarctica lipase A (CaLA), the enantioselectivity was poor.
Nitromethane and acetonitrile likewise proved to be unsuitable for this reaction. No adequate resolution of 2c upon acylation with vinyl acetate could be obtained in any solvent.
Table 2.2 Resolution of 1-nitro-2-pentanol in the presence of CaLB; effect of the acyl donor and the solventa
Solvent Succinic anhydride Vinyl acetate conv (%) eeS (%) eeP (%) E conv (%) eeS (%) eeP (%) E MeNO2 <1 0 1 - 3 1 45 3 ACN 8 2 100 2 nda nd nd nd DME 4 3 76 7 37 38 52 5 TBME 42 70 95 100 46 41 35 3 DIPE 54 92 93 82 37 15 25 2
Reaction conditions: 1 ml diisopropyl ether, 100 mM 1-nitro-2-pentanol, 100 mM acyl donor, 25 mg Novozym 435
a Not determined
Substrate specificity and enantiodiscrimination
A range of nitro alcohols, resulting from aldol reaction of aliphatic (1a-c) and aromatic (1d-f) aldehydes with nitromethane, was studied in the lipase-mediated resolution with succinic anhydride. The selectivity of Novozym 435 for these substrates follows the Kazlauskas rule15 (Figure 2.2).
H OH
M L
Figure 2.2 Steric model for the preferentially converted enantiomer of secondary alcohols by a lipase
Lipase catalysed Resolution of Nitro-aldol Adducts
26
Such effects would become visible when both enantiotopic groups are similar in size. We found, however, that the enantiomer discrimination of 2b (R=Et) was much lower that that of 2a or 2c (R=Me or Pr, respectively, see Table 3). We conclude from these results that enantiodiscrimination of these compounds by CaLB is dominated by steric interactions and that electrostatic effects are at best minor.
The longer alkyl- (2c) and the phenyl alkyl substituted (2d-f) substrates all show (S)-selectivity in accordance with the predictive model shown in Figure 2.2. Surprisingly, there was no reaction observed with 3-phenyl-1-nitro-2-propanol (2e) when succinic anhydride is used as acyl donor. We have no plausible explanation for this observation.
During the resolution of the aromatic nitro alcohols some spontaneous elimination of carboxylic acid into the corresponding nitroalkene was observed. Elimination of acetic acid occurred more readily than that of succinic acid. No detectable alkene formation from the aliphatic nitroalcohols and their esters was observed under the reaction conditions.
Overall, much higher enantioselectivities were found in the resolution of 1-nitro-2-alkanols when succinic anhydride was used as the donor compared with vinyl acetate.
Conclusion
NO2 O O (S)-3c NO2 O O (R)-3b NO2 O O (R)-3a OH O OH O OH O NO2 O O (S)-3f NO2 O O (S)-3e NO2 O O (S)-3d OH O OH O OH O R O R NO2 OH R NO2 OH R NO2 O O OH O + + Lipase Succinic anhydride H3C NO2
Figure 2.3 Preferentially formed enantiomers of 1-nitro-2-alkanols by Novozym 435
Table 2.3 Resolution of 1-nitro-2-alkanols with succinic anhydride and vinyl acetate in
diisopropyl ethera
Succinic anhydride Vinyl acetate R Conv (%, 24 h) ees (%) eep (%) E Conv (%, 24h) ees (%) eep (%) E CH3 39a 57 67 28 (R) 30 2 5 1 (R) C2H5 47a 44 43 4 (R) 72 1 ndc 1c(R) C3H7 54 92 93 82 (S) 37 15 25 2 (S) C6H5 4d 3 75 7 (S) 13d 13 100 20c(S) (C6H5)CH2 nr - - - 53d 34 ndb 2c(S) (C6H5)C2H4 42 71 97 96 (S) 56d 21 100 2c(S) Reaction conditions: 1 ml diisopropyl ether, 100 mM substrate, 100 mM succinic anhydride, 25 mg Novozym 435
a Reaction finished after 5h c Not determined
c Calculated only with ees.
Lipase catalysed Resolution of Nitro-aldol Adducts
28
Experimental part- Material and methods
Instruments and materials
HPLC analyses were performed on a Chiralcel OD column with a flow of 0.6 ml/min and an eluent consisting of hexane-isopropyl alcohol-trifluoroacetic acid (95:5:0.1) for the aliphatic nitro alkanols and hexane-isopropyl alcohol-trifluoroacetic acid (80:20:0.1) for the aromatic nitro alkanols. Detection was performed with a Waters 486 UV detector at 215 nm.
All chemicals were of analytical purity and obtained from Sigma-Aldrich. The lipase from Candida rugosa was obtained from Sigma-Aldrich. Candida
lypolitica lipase was bought from Fluka. Novozym 435 (Candida antarctica
lipase B on Lewatit E), SP524 (Rhizomucor miehei lipase), SP526 (Candida
antarctica lipase A) were kindly donated by Novozymes. Cross-linked enzyme
crystals (CLECs) were donated by Altus Biologics; CaLB CLEA was donated by CLEA Technologies. The 1-nitro-2-alkanols were synthesized as described below.
Methods
General synthesis of aliphatic nitro alcohols 2a-c
Aldehyde 1a-c (100 mmol) and KOH (1 ml, 1M solution) were added to nitromethane (10 ml) at 0-5 °C. After 1 h the mixture was brought to room temperature and left for another 2-4 h. The reactions were followed by TLC using ether-petroleum ether 3:2 as the eluent. Dichloromethane was added and this mixture was washed successively with 5% aqueous HCl and saturated aqueous NaHCO3. The organic phase was dried with MgSO4 and
1-Nitro-2-propanol 2a Isolated yield 46%. 300 MHz 1H NMR (CDCl3) δ (ppm): 1.3 (3H, d), 3.2 (1H, s), 4.4(2H, m), 4.5 (1H, m). 75 MHz 13C NMR (CDCl3) δ (ppm): 19.8, 65.0, 81.6 1-Nitro-2-butanol 2b Isolated yield 48%. 75 MHz 13C NMR (CDCl 3) δ (ppm): 5.6, 26.9, 70.1, 80.5 1-Nitro-2-pentanol 2c
The reaction was carried out on 1M scale. Isolated yield 69%. 300 MHz 1H NMR (CDCl3) δ (ppm): 1.0 (3H, t), 1.6 (2H, m), 3.0(1H, s), 4.3 (1H, m), 4.5
(2H, m).
General synthesis of aromatic nitro alcohols 2d-f
Aldehyde 1d-f (100 mmol) and KOH (1 ml, 1M solution) were added to nitromethane (20 ml) at 0-5°C. After 1 h the mixture was brought to room temperature and left stirring for another 2-4 h. The reactions were followed by TLC using ether-petroleum ether 1:1 as the eluent. The pH was adjusted to 7 and the excess of nitromethane was evaporated. Ether was added and washed successively with acidic water, saturated NaHCO3 solution and water.
The organic phase was dried with MgSO4 and the solvent evaporated to yield
the crude product.
2-Phenyl-1-nitro-2-ethanol 2d
The crude product was purified by distillation giving a yellow liquid. Isolated yield 57%. 300 MHz 1H NMR (CDCl
3) δ (ppm): 3.2 (1H, s), 4.5 (2H, m), 5.4
(1H, q), 7.3 (5H, m). 75 MHz 13C NMR (CDCl3) δ (ppm): 70.9, 81.1, 125.9,
128.9, 129, 138.2.
3-Phenyl-1-nitro-2-propanol 2e
Lipase catalysed Resolution of Nitro-aldol Adducts
30
(1H, q), 7.3 (5H, m). 75 MHz 13C NMR (CDCl3) δ (ppm): 40.4, 69.5, 79.7,
127.3, 128.9, 129.4, 135.9.
4-Phenyl-1-nitro-2-butanol 2f
The product was recrystallised from diisopropyl ether resulting in a white solid. Isolated yield 56%. 300 MHz 1H NMR (CDCl3) δ (ppm): 1.8 (2H, m), 2.6 (1H,
d), 2.85 (2H, m), 4.35 (1H, m), 4.45 (2H, m), 7.3 (5H, m). 75 MHz 13C NMR (CDCl3) δ (ppm): 31.3, 35.1, 67.7, 80.5, 126.3, 128.4, 128.6, 140.6.
Lipase-catalysed resolutions
The lipase-catalysed acylation reactions were performed in 1 ml solvent at room temperature, with 100 mM nitro alcohol, an equivalent amount of acyl donor and 25 mg of the various enzyme preparations. When CLECs or CLEAs were used 5 mg was added instead to ensure the use of the same amount of active protein. Except for the enzyme screening, all reactions were carried out with Novozym 435. Trimethoxybenzene (2.5 g/l) was used as the internal standard. The reactions were monitored by chiral HPLC.
References
1 Machajewski, D.; Wong, C.H. Angew. Chem. Int. Ed. Engl. 2000, 39, 1352-1374 2 Itayama, T. Tetrahedron 1996, 52, 6139-6148
3 Nakamura, K.; Kitayama, T.; Inoue, Y.; Ono, A. Tetrahedron 1990, 46, 7471-7481 4 Bornscheuer, U.T.; Kazlauskas, R.J. Hydrolases in Organic Synthesis; Wiley-VCH:
Weinheim, 1999
5 Jaeger, K.; Eggert, T. Curr. Opin. Biotechnol 2002, 13, 390-397
6 Schmid, A.; Hollman, F.; Park, J.B.; Bühler, B. Curr. Opin. Biotechnol. 2002, 13, 359-366
8 Morgan, B. et al. Enzymes in non-aqueous solvents, Vulfson, E.N.; Halling, P.J.; Holland, H.L., eds, Humana press: New York, 2001, p. 444-451
9 Gutman, A.L.; Brenner, D.; Boltanski, A. Tetrahedron: Asymmetry, 1993, 4, 839-844 10 Hyatt, J.A.; Skelton, C. Tetrahedron: Asymmetry, 1997, 8(4), 523-526
11 Fiaud, J.; Gil, R.; Legros, J.; Aribi-Zouioueche, L.; König, W.A. Tetrahedron Lett. 1992, 33, 6967-6970
12 Nakamura, K.; Takenaka, K.; Ohno, A. Tetrahedron: Asymmetry 1998, 9, 4429-4439 13 Terao, Y.; Tsuij, K.; Murata, M.; Achiwa, K.; Nishio, T.; Watanabe, N.; Seto, K. Chem.
Pharm. Bull. 1989, 37, 1653-1655
14 Patel, R.N.; Banerjee, A.; Nanduri, V.; Goswami, A.; Comezoglu, F.T. J.Am.Oil Chem. Soc. 2000, 77, 1015-1019
15 Kazlauskas, R.J.; Weissfloch, A.N.E.; Rappaport, A.T.; Cuccie, L.A. J. Org. Chem.
3
Optimizing the
Deracemisation of Nitro
aldol Adducts
Abstract
Optimizing the Deracemisation of Nitro-aldol Adducts
34
Introduction
In Chapter 2, a viable resolution method for β-nitro alcohols has been demonstrated, employing esterification with succinic anhydride catalysed by Novozym 435.1 The acyl donor was found to have a profound influence on the success of this resolution (Figure 3.1). A 50 times higher E value could be achieved when succinic anhydride (E 96) was used as acyl donor instead of the commonly employed vinyl acetate (E 2).
NO2 OH Novozyme 435 Succinic anhydride + NO2 OH NO2 O O OH O
Figure 3.1 Lipase catalysed resolution of 4-phenyl-1-nitro-2-butanol with succinic anhydride
To overcome the limitation of a 50% maximum yield typical for a classical kinetic resolution the possibility of a dynamic kinetic resolution (DKR) for β-nitro alcohols was investigated. A dynamic kinetic resolution combines the enantioselective resolution with an in-situ racemisation of the slow reacting enantiomer, making full conversion of a racemic substrate into one single enantiomer possible.2 Various methods have been applied and studied to achieve this goal. A recent promising method for secondary alcohols combines an enzymatic resolution with a ruthenium-complex catalysed racemisation of the remaining enantiomer (Figure 1.2).3,4,5 Another approach yielding the same result is to keep the starting material racemic by using a fast reversible reaction via a prochiral precursor. The latter approach has been effective in the synthesis of enantiopure cyanohydrins, by using the equilibrium of the hydrocyanation of carbonyl compounds.6,7,8 Aldol reactions can be used in a similar fashion when the reversibility of the aldol reaction is used to racemise the alcohol (Figure 3.2).
unstable. Therefore it is impossible to isolate it and an alternative procedure is required. NO2 O O (R) OH O Novozym 435 Succinic anhydride (S) O Me-NO2 + NO2 OH NO2 OH Base
Figure 3.2 A dynamic kinetic resolution using the equilibrium of the aldol reaction
A resolution can be performed in subtractive manner, isolating the fast reacting enantiomer as the ester after resolution or in an attractive manner, isolating the slow reacting enantiomer of the substrate.9 The latter method is more preferable because it consists of fewer steps to gain the pure homochiral alcohol. Besides it is easier to gain high ee even with moderate enzymatic selectivity by controling how far the reaction is allowed to proceed. The previous described in-situ racemisation methods for enhancing the efficiency of a resolution need to be combined with a subtractive resolution due to the fact that the alcohol is used for racemisation.
Optimizing the Deracemisation of Nitro-aldol Adducts 36 NO2 O O (R) OH O Novozym 435 Succinic anhydride (S) NO2 OH NO2 OH Base + NO2 H+ H2O
Figure 3.3 Reaction scheme for the subtractive resolution and combined racemisation of β-nitro alcohols
In this chapter the feasibility of the in situ and ex situ racemisation methods to improve the deracemisation of β-nitro alcohols is reported.
Results and discussion
Dynamic kinetic resolution and instability of the ester
The kinetic resolution of the test substrate 4-phenyl-1nitro-2-butanol has been studied and was reported previously in Chapter 2. All trials for racemising the slow-reacting enantiomer of the alcohol, by using the equilibrium of the aldol reaction, were unsuccessful. Catalytic amounts of Amberlyst A21, NaOH, pyridine or triethylamine were used as aldol reaction catalysts for the purpose of racemisation during the resolution reaction. Instead of gaining better yields in ester formation and a racemic alcohol throughout the full course of the resolution, the main product was the corresponding alkene. The instability of the ester can be ascribed to the acidic α-proton next to the nitro-group, which enables elimination via a cyclic mechanism (Figure 3.4). When elimination occurs according to this mechanism it will result in an alkene with cis configuration. On the other hand, dissociation via standard E1 elimination will favour the trans configuration. To verify the hypothesis that instability of the ester can be ascribed to the acidic α-proton next to the nitro-group, the stereochemistry around the double bond has been elucidated on the basis of its 1H NMR spectrum and 1H NOESY NMR. Although in the 1H NMR ja-b = 13.5
found in the NOESY experiment for Ha and Hb (Figure 3.4). This is strong
evidence for a cis configuration of the alkene and supports the proposed mechanism of elimination. O H O NO2 R
R: Restgroup of the acyl donor
O R OH NO2 Ha Hb Hc Hd He He
Figure 3.4 Dissociation mechanism of the esters of β-nitro alcohols
Recycling of the unstable ester
To overcome the problem of product instability the resolution is performed subtractively, isolating the slow reacting enantiomer of the alcohol. Following the reaction scheme shown in Figure 3.3 the ester that is formed from the fast reacting enantiomer can than easily be converted in the cis-alkene with Amberlyst A21. The main challenge is the subsequent hydration of this alkene to yield the racemic nitro alcohol. A standard procedure to achieve water addition to a double bond is by using concentrated sulfuric acid. The alkene reacts with concentrated sulfuric acid to give an alkyl hydrogen sulfate. The alkyl hydrogen sulfate can be converted by SN1 substitution to an alcohol by
boiling in water. Due to the electron withdrawing effect of the nitro group present in the alkene, and therefore a more stable carbocation on the β-position than on the α-β-position during the formation of the alkyl hydrogen sulfate, this will lead to the correct position of the alcohol.
Table 3.1 Hydration of 4-phenyl-1-nitrobut-1-ene
ConcentrationH2SO4 Conversion 1 h (%) Conversion 2 h (%)
5 M 17 24
6 M 13 20
7 M 6 21
8 M 1 22
Optimizing the Deracemisation of Nitro-aldol Adducts
38
Initially (1h of reaction) low concentrations of sulphuric acid seem beneficial, but final conversion after 2h is similar for all conditions (Table 3.1). Due to the hydration being exothermic together with a decrease in number of molecules, the equilibrium of this reaction is thermodynamically not favourable for hydration. Low temperatures would be beneficial, but due to the low reactivity of the alkene more severe reaction conditions are often required.10 Lower and higher temperatures than 100°C (75°C, 130°C) did not show any improvement and conversions lowered drastically into the few percent range.
A cycle combining the resolution and the hydration with intermediate alcohol isolation (Figure 3.5) resulted in an isolated total yield of 48 %. Substantial enantiomeric enrichment of the produced alcohol was achieved during crystallisation raising the ee from 70 % to >99 %.
Figure 3.5 Integrated deracemisation cycle
Dynamic kinetic resolution of
α
-proton deficient substrates
To study the intrinsic possibility to use the reversible Henry reaction for designing a dynamic kinetic resolution as shown in Figure 3.2, substrates without the acidic α-protons where studied. The esters that will be formed during lipase-catalysed resolution will be much more stable because they cannot dissociate according to the mechanism shown in Figure 3.4.
R NO2 OH
Novozym 435
Candida antarctica lipase B CLEA Candida antarctica lipase A Pseudomonas cepacia lipase Candida rugosa lipase Acylase 1 (esterase) Hydrolase Acyl donor R NO2 O R' O R= CH3 C2H5 C6H5 -C2H4 R'= C2H4COOH CH3 Hydrolases
Figure 3.6 Resolution experiments with α-proton deficient substrates
Ethyl esters of the three α-proton deficient substrates proved to be much more stable indeed under the conditions needed for the Henry reaction (Figure 3.6). Unfortunately these substrates, of which the α-protons were replaced by methyl groups, were not accepted by any of the lipases that were tested with succinic anhydride and vinylacetate as acyl donor. Also the esterase in Acylase 1 did not show any activity. The presence of a quaternary carbon atom directly next to the chiral centre apposes too much steric constrains for these substrates to be able to bind effectively in the active site.
Conclusion
Optimizing the Deracemisation of Nitro-aldol Adducts
40
conditions. The more stable α-proton deficient substrates, to prove the principle, were not accepted as substrate by the lipases tested.
The yield of alkene hydration, for recycling the dissociated ester towards racemic alcohol is too low to be efficient.
Experimental part: Materials and Methods
Instruments and materials
All chemicals were of analytical purity and obtained from Sigma-Aldrich. Novozym 435 (Candida antarctica lipase B on Lewatit E) was kindly donated by Novozymes. CaLB-CLEA was made available by CLEA Technologies. The nitro alcohols were synthesized via Henry reaction as described below. HPLC analyses were performed on a Chiralcel OD column with a flow of 0.6 ml/min and an eluent consisting of hexane-isopropyl alcohol-trifluoroacetic acid (80:20:0.1). Detection was performed with a Waters 486 UV detector at 215 nm.
Methods
Synthesis of 4-phenyl-1-nitro-2-butanol
3-Phenylpropionaldehyde (1 M), nitromethane (2 M) and KOH (1 ml, 5M solution) were added to 1,2-dimetyhoxyethane (200 ml) at 0-5°C. After one hour the mixture was brought to room temperature and left stirring for another 2-4 hours. The reaction was followed by TLC using ether-petroleum ether 1:1 as the eluent. After the reaction was finished the pH was neutralized and the excess of nitromethane and solvent were evaporated. Ether was added and washed sequentially with acidic water, saturated NaHCO3 solution and water.
The organic phase was dried with MgSO4 and the solvent evaporated to yield
(ppm): 1.8 (2H, m), 2.6 (1H, d), 2.85 (2H, m), 4.35 (1H, m), 4.45 (2H, m), 7.3 (5H, m). 75 MHz 13C NMR (CDCl3) δ (ppm): 31.3, 35.1, 67.7, 80.5, 126.3,
128.4, 128.6, 140.6.
Synthesis of 4-phenyl-1-nitrobut-1-ene
4-phenyl-1-nitro-2-butanol (20g) and acetic anhydride (9.6 ml) were dissolved in 20 ml of ether. Pyridine (9.2 ml) was added and the reaction mixture was stirred for 4 h at room temperature. Ester formation was followed by HPLC and TLC using ether-petroleum ether 1:1 as the eluent. The reaction mixture was washed sequentially with acidic water, saturated NaHCO3 solution and
water. The organic phase was dried with MgSO4. The formation of
4-phenyl-1-nitrobut-1-ene was induced by refluxing the resulting solution in ether for 2 h. After evaporation of the solvent the reaction mixture was distilled and the product was collected at 140 °C and 5⋅10-3 mbar. Isolated yield 61 %. 300
MHz 1H NMR (CDCl3) δ (ppm): 2.6 (2H, m), 2.9 (2H, t), 6.9 (1H, m), 7.1-7.3
(5H, m). The phase-sensitive NOESY 1H NMR spectra were obtained with 512 fn1 increments and 2048 data points. 4 free induction decays were collected for each value of fn1. The mixing time used was 0.1 s.
Synthesis of 1-phenyl-4-methyl-4-nitro-3-pentanol
3-Phenylpropionaldehyde (1.3 ml, 10 mmol), 2-nitropropane (1.8 ml, 20 mmol) and Amberlyst A21 in OH- form (1 g) were added to ethanol (100 %, 25 ml) at 0-5°C. After one hour the mixture was brought to room temperature and left stirring for another 20 hours. The reaction was followed by TLC using ether-petroleum ether 1:1 as the eluent. After the reaction was finished the Amberlyst A21 was filtered and washed with ethanol. The organic filtrate was washed sequentially with acidic water, saturated NaHCO3 solution and water.
The organic phase was dried with MgSO4 and the solvent and the excess of
2-nitropropane evaporated to yield the crude product. The product was further purified by destillation and collected at 80 °C and 4.10-2 mbar. Isolated yield
Optimizing the Deracemisation of Nitro-aldol Adducts
42
Synthesis of 2-methyl-2-nitro-3-pentanol
Propionaldehyde (7.3 ml, 100 mmol), 2-nitropropane (9 ml, 100 mmol) and Amberlyst A21in OH- form (4 g) were added to dichloromethane (30 ml) at 0-5°C. After one hour the mixture was brought to room temperature and left stirring for another 20 hours. The reaction was followed by TLC using ether-petroleum ether 1:1 as the eluent. After the reaction was finished the Amberlyst A21 was filtered and washed with dichloromethane. The organic filtrate was washed sequentially with acidic water, saturated NaHCO3 solution
and water. The organic phase was dried with MgSO4 and the solvent and the
excess of 2-nitropropane evaporated to yield the crude product. The product was further purified by distillation. Isolated yield 10 %.
Synthesis of 2-dimethyl-2-nitro-3-hexanol
Butyraldehyde (4.4 ml, 50 mmol), 2-nitropropane (4.5 ml, 50 mmol) and Amberlist A21 in OH- form (2 g) were added to ethanol (100%, 25 ml) at
0-5°C. After one hour the mixture was brought to room temperature and left stirring for another 20 hours. The reaction was followed by TLC using ether-petroleum ether 1:1 as the eluent. After the reaction was finished the Amberlyst A21 was filtered and washed with dichloromethane. The organic filtrate was washed sequentially with acidic water, saturated NaHCO3 solution
and water. The organic phase was dried with MgSO4 and the solvent and the
excess of 2-nitropropane evaporated to yield the crude product. The product was further purified by distillation. Isolated yield 26 %.
Lipase-catalysed resolutions
The lipase-catalysed acylation was performed in diisopropylether (1 ml) as solvent at room temperature, with 100 mM nitro alcohol, an equivalent amount of succinic anhydride and 25 mg of Novozyme 435. Trimethoxybenzene (2.5 g/l) was used as the internal standard. The reactions were monitored by chiral HPLC. For the dimethylated nitro alcohols Novozym 435 (25 mg),
(CaLA, 25 mg), Candida rugosa lipase (CR, 25 mg), Candida antarctica Lipase B CLEA (CaLB-CLEA, 5 mg) were screened.
Conversion of the ester into racemic alcohol
The resulting mixture from the resolution was treated with Amberlist A21(OH -form, 100 mg) for 1 h to completely convert the succinate ester of 4-phenyl-1-nitro-2-butanol into the 4-phenyl-1-nitrobut-1-ene. After separation of the solids diisopropyl ether was evaporated and a sulfuric acid solution (various concentrations, amount sufficient to get nitro alcohol concentration to be 1 M) was added. After incubation for 30 minutes at room temperature the mixture was heated till 80 °C for 2 hours and after cooling, and neutralizing the pH with NaHCO3 extracted with diisopropylether. Analysis of the samples was
performed by chiral HPLC.
References
1 Sorgedrager, M.J.; Malpique, R.; van Rantwijk, F.; Sheldon, R.A.Tertraherdron: Asym.
2004, 15, 1295-1299
2 Pellissier, H.; Tetrahedron, 2003, 59, 8291-8327
3 Kim, M.J.; Ahn, Y.; Park, J. Curr. Opin. Biotechnol. 2002, 13, 578-587
4 Huarta, F.F.; Minidis, A.B.E.; Bäckvall, J.E. Chem. Soc. Rev. 2001, 30, 321-331 5 Pàmies, O.; Bäckvall, J.E. Curr. Opin. Biotechnol. 2004, 14, 407-413
6 Inagaki, M.; Hiratake, J.; Nishioka, T; Oda, J. J. Org. Chem. 1992, 57, 5643-5649 7 Li, Y.X.; Straathof, J.J.; Hanefeld, U. Tetrahedron: Asym. 2002, 13, 739-743
8 Paizs, C.; Tosa, M.; Majdik, C.; Tähtinen, P.; Irimie, F.D.; Kanerva, L.T. Tetrahedron: Asym. 2002, 14, 619-627
Part II
4
Asymmetric Reduction with
Candida magnoliae
Ketoreductase S1
Abstract
Asymmetric Reduction with Candida magnoliae Ketoreductase S1
48
Introduction
Biocatalytic reduction of ketones is a convenient route towards homochiral secondary alcohols. Especially the synthesis of the enantiopure side chain of the statins (Chapter 1) has gained a remarkable amount of interest over the past years. Various synthetic approaches have been studied (Figure 1.10) of which the biocatalytic reduction of the corresponding β-ketoesters and β,δ-diketoesters is a promising method. This approach has been the target of various research projects in which the reduction of the halogenated ethyl-4-chloroacetoacetate (1a) gained most interest. Besides being interesting for the synthesis of the statins, this compound is also an intermediate in the synthesis of carnitine and 1,4-dihydropyridine type β-blockers.1 Baker’s yeast reductions
have been performed with these type of ketoesters yielding the (S)-enantiomer.2 The stereochemistry of these reductions with Saccharomyces cerevisiae could be controlled by the size of the ester group.2,3 However the cheap and readily available baker’s yeast did not always give the right configuration or optical purity. In the case of ethyl-4-chloroacetoacetate (1a) the major product had the desired (S)-configuration but only with 55% ee.2
Over 400 other microorganisms have been screened for their ability to reduce 4-chloroacetoacetate ethyl ester.4 The yeast Candida magnoliae performed best in this reduction. Heat-treated cells produced 90 g/l (S)-ethyl-4-chloro-3-hydroxybutanoate ((S)-2a) with 99% ee in the presence of a cofactor regeneration system based on glucose dehydrogenase (Figure 4.1).5 The enzyme involved in this latter reaction, the NADPH dependent ketoreductase S1, has been isolated, purified and characterised.6 Subsequently it has been overexpressed in E. coli.7 A combined overexpression of the genes encoding for C. magnoliae ketoreductase S1 and the glucose dehydrogenase of
Ketoreductase S1 is originally NADPH dependent. Regeneration systems for this cofactor are less attractive than for NADH, due to the costs of the cofactor and the lower stability of NADPH. There is also a paucity of enzymes that mediate the reduction of NADPH by suitable sacrificial reductants. The popular regeneration system for NADH, based on formate dehydrogenase (FDH), produces CO2 gas, while in the regeneration system for NADPH
gluconate is produced (Figure 4.1). In the latter case a more laborious downstream processing is required. Therefore, attempts have been made to alter the cofactor specificity of ketoreductase S1.9,10 Computer modelling of homology models for S1 have been used to identify amino acid residues in the adenosine binding pocket that could be suitable targets for site-directed mutagenesis. The most efficient mutant in those studies produced 163 g/l (S)-2a with 99% ee using NADH as cofactor combined with a regeneration system based on formate dehydrogenase.9
O R3 O O R2 O R3 O R2 NADPH NADP+ NADH NAD+ Glucose Gluconate HCO2H CO2 GDH FDH OH 1a-d 2a-d R1 C.m. KRED a: R1=Cl; R2=H; R3=C2H5 b: R1=H; R2=H; R3=C2H5 d: R1=H; R2=CH3; R3=C2H5 c: R1=H; R2=H; R3=C(CH3)3 R1
Figure 4.1 Asymmetric reduction of β-ketoesters with common used cofactor regeneration
systems
The relative substrate specificity and enantioselectivity of C. magnoliae ketoreductase S1 was investigated with a number of substituted β-ketoesters (Table 4.1).11,12 The R1 group has a huge influence on the activity. Comparing
Asymmetric Reduction with Candida magnoliae Ketoreductase S1
50
activity. The presence of electron-withdrawing substituents at R1 apparently
activates the carbonyl group for hydride attack.
The enzyme also seems to have a very pronounced preference for ethyl esters. The activity with the smaller methyl ester of 4-chloroacetoacetate (1f) was only 11% of that of the ethyl ester (1a). The much larger and more hydrophobic octyl ester was also much less reactive (36% relative to 1a) and the bulky tert-butyl ester was reported to be unreactive.11,12
Table 4.1 Enantioselective reduction of various β-ketoesters
R1 R2 R3 Relative activity (%) ee (%) Ref 1a Cl H C2H5 100 n.db 11 1b H H C2H5 7 >99 R 12 1c H H C(CH3)3 0 n.d 12 1d H CH3 C2H5 14 (2S,3R) 19;(2S,3S) 78a (2R,3R) 3 ;(2R,3S) <1 12 1e H Cl C2H5 11 (2S,3R) 2 ;(2S,3S) 56 a (2R,3R) 41;(2R,3S) <1 12 1f Cl H CH3 11 n.d 11 1g Cl H C8H17 36 n.d 11 1h Br H C2H5 72 >99 S 12 1i I H C2H5 16 >99 S 12 1j N3 H C2H5 0 n.d 12 1k OH H C2H5 80 >99 S 12 1l C6H5CH2O H C2H5 21 21 S 12 1m C2H5 H C2H5 0.5 n.d 12
a Product distribution in %; preferential syn product = (2S,3R); preferential anti product = (2S,3S) b Not determined
S1 can tolerate ethyl acetate, butyl acetate and diisopropyl ether as co-solvents in a biphasic system.10,13
We have been employing directed evolution using DNA shuffling technology, to increase the activity of the ketoreductase S1 from Candida
magnoliae for ethyl-4-chloroacetoacetate (1a).13 In this chapter we report the behaviour of various mutants selected from 10 rounds of shuffling according to their activity with 1a, as well as related β-ketoesters (1b-1d). The influence of the cofactor and of co solvents has been investigated.
Results and discussion
The variants that were studied originated from various generations of directed evolution. CmKr1, CmKr2A, CmKr2B, CmKr4, CmKr8, CmKr9, CmKr10 are selected variants of Candida magnoliae ketoreductase S1 from the library of round 1, 2, 4, 8, 9 and 10 of DNA shuffling respectively.
Reduction of
β
-ketoesters
The first substrate (1a) is the substrate the enzyme is evolved for. A 16-fold increase in activity was observed in 10 rounds of DNA shuffling without any loss of enantioselectivity. Only the variant CmKr2A showed a slight decrease in enantioselectivity (Table 4.2, entry 2). This variant was from a library where it was attempted to incorporate a change of cofactor dependency towards NADH into the development of this enzyme. Although the data shown in Table 4.2 is for NADPH, the same decrease in ee was found using NADH as cofactor. Due to the decreased enantioselectivity this strategy was abandoned and further evolution was continued from the library of round 1 with NADPH as cofactor. The variant CmKr2B is part of the resulting second generation library. The later libraries are based on that one.
Asymmetric Reduction with Candida magnoliae Ketoreductase S1
52
improvement (5-fold) was found with CmKr2A, which originated from the library that was developed co-targeting NADH dependency.
The activity decreased drastically when replacing the ethyl ester with the more bulky tert-butyl one (1c). Although in previous studies no activity was found with this substrate in the wild type enzyme (Table 4.1), we found low activities in all variants with in general high enantioselectivity (ee >99%). Since at the moment of writing no crystallographic data is available for this enzyme, no structural explanation can be given. We suggest that steric congestion due to the tert-butyl group made it difficult for the ketoester to bind in the active site with the keto group in the right position for hydride transfer.
Table 4.2 Asymmetric reduction of β-ketoesters catalysed by variants from 10 rounds of
directed evolution on Candida magnoliae ketoreductase S1
Cl O OH O 2a O OH O 2b O OH O 2c Enzyme Vini µmol⋅min-1⋅ g-1 ee (%) Vini µmol⋅min-1⋅ g-1 ee (%) Vini µmol⋅min-1⋅ g-1 ee (%) CmKr1 481 >99 (S) 43 >99 (R) 92 >99 (R) CmKr2A 1865 98.5 (S) 311 >99 (R) 17 >99 (R) CmKr2B 551 >99 (S) 12 >99 (R) 11 73 (R) CmKr4 1700 >99 (S) 149 >99 (R) 20 98 (R) CmKr8 5397 >99 (S) 229 >99 (R) 53 >99 (R) CmKr9 4615 >99 (S) 94 >99 (R) 24 >99 (R) CmKr10 7763 >99 (S) 214 >99 (R) 41 >99 (R)
Activity measurement conditions: 2 ml Tris-HCl buffer (50 mM, pH 7.6), 0.16 mM NADPH, 10 mM substrate, 0.1 mg enzyme
Enantioselectivity measurement conditions: 0.2 ml Tris-HCl buffer (50 mM, pH 7.6), 11 mM NADPH, 10 mM substrate, 0.2 mg enzyme
The stereoselectivity of reduction of the model substrates 1a-1c with C.
magnoliae ketoreductase S1 was not as predicted by Prelog’s rule for
The mutations that resulted in the very considerable rate improvement in the reduction of 1a did not lead to similar improvements with the related compounds 1b-1d. Neither did these narrow down the substrate specificity, as the activity for the non-chlorinated substrates (1b-1d) did not decrease through the generations of evolution. The enantioselectivity remained high in all variants for all model substrates.
Reduction of
α
substituted ketoesters
Ethyl-3-oxobutanoates bearing a substituent on C-2 have a chiral center at position 2 and a prochiral carbon at position 3. It is interesting to study how these substrates are handled by ketoreductase S1 from Candida magnoliae. Baker’s yeast,15,16 Geotrichum candidum17and Candida magnoliae12 are known to reduce these substrates. Whole cells of baker’s yeast reduce the ethyl-2-methyl-3-oxobutanoate (1d) exclusively yielding the (2R,3S)-configuration,19 whereas the 2-chloro substituted compound was reduced into
a 1:1 mixture of (2S,3S) and (2R,3S) product. For a number of 2-substituted ethyl-3-oxobutanoates whole cells of baker’s yeast as well as a large number of its purified reductases give exclusively (3S) products.16,18,19 The selectivity for the configuration on C-2 varies and is dependent on the substrate and on which particular enzyme is used.
Wild type Candida magnoliae ketoreductase S1 is reported to reduce these substrates with relatively low selectivities for both the configuration at C-2 and C-3.12 In the case of 1d the configuration of the main product was (2S,3S). Minor amounts of (2S,3R) and (2R,3S) were formed and hardly any (2R,3R) product could be detected. With a chlorine as substituent on C-2 the major products were (2S,3S) and (2R,3R) with only a slight excess of (2S,3S).
