Enthalpy and Entropy in Enzyme Catalysis A Study of Lipase Enantioselectivity
Jenny Ottosson
Department of Biotechnology
Royal Institute of Technology
Stockholm, Sweden 2001
Cover: Positions of the heavy atoms of (R)-3-hexanol in a 500 ps molecular dynamics
simulation. The oxygen atom is colored black. The chiral carbon is colored yellow and the
other carbons red, blue and green.
© Jenny Ottosson
Printed at Universitetsservice US AB, Stockholm, Sweden, 2001
ISBN 91-7283-157-X
Jenny Ottosson (2001): Enthalpy and entropy in enzyme catalysis - A study of lipase
enantioselectivity.
Department of Biotechnology, Royal Institute of Technology, KTH, Stockholm, Sweden.
ISBN 91-7283-157-X
Abstract
Biocatalysis has become a popular technique in organic synthesis due to high activity
and selectivity of enzyme catalyzed reactions. Enantioselectivity is a particularly attractive
enzyme property, which is utilized for the production of enantiopure substances.
Determination of the temperature dependence of enzyme enantioselectivity allows for
thermodynamic analyses that reveal the contribution of differential activation enthalpy,
DR-SDH‡, and entropy, DR-SDS‡. In the present investigation the influence of substrate structure,
variations on enzyme structure and of reaction media on the enantioselectivity of Candida
antarctica lipase B has been studied.
The contribution of enthalpy, DR-SDH‡, and entropy, TDR-SDS‡, to the differential free
energy, DR-SDG‡, of kinetic resolutions of sec-alcohols were of similar magnitude. Generally
the two terms were counteracting, meaning that the enantiomer favored by enthalpy was
disfavored by entropy. 3-Hexanol was an exception where the preferred enantiomer was
favored both by enthalpy and by entropy. Resolution of 1-bromo-2-butanol revealed
non-steric interactions to influence both DR-SDH‡ and DR-SDS‡. Molecular modeling of the
spatial freedom of the enzyme-substrate transition state indicated correlation to the transition
state entropy. The acyl chain length was shown to affect enantioselectivity in
transesterifications of a sec-alcohol.
Point mutations in the active site were found to decrease or increase enantioselectivity.
The changes were caused by partly compensatory changes in both DR-SDH‡ and DR-SDS‡.
Studies on single and double mutation variants showed that the observed changes were not
additive.
Enantioselectivity was strongly affected by the reaction media. Transesterifications of
a sec-alcohol catalyzed by Candida antarctica lipase B was studied in eight liquid organic
solvents and supercritical carbon dioxide. A correlation of enantioselectivity and the
molecular volume of the solvent was found.
Differential activation enthalpy, DR-SDH‡, and entropy, DR-SDS‡, display a
compensatory nature. However this compensation is not perfect, which allows for
modifications of enantioselectivity. The components of the thermodynamic parameters are
highly complex and interdependent but if their roles are elucidated rational design of
enantioselective enzymatic processes may be possible.
Keywords: Biocatalysis, enzyme catalysis, Candida antarctica lipase B, enantioselectivity,
enthalpy, entropy, CALB, enantiomeric ratio
© Jenny Ottosson, 2001
Sammanfattning
Biokatalys har blivit en populär teknik inom organisk syntes på grund av
enzymkatalyserade reaktioners höga aktivitet och selektivitet. Enantioselektivitet är en
speciellt attraktiv egenskap hos enzymer som utnyttjas för produktion av enantiomert rena
substanser. Enantioselektivitetens temperaturberoende möjliggör termodynamisk analys av
bidragen från differentiell aktiveringsentalpi, ∆R-S∆H‡, och aktiveringsentropi, ∆R-S∆S‡. I
denna undersökning har inverkan av substratstruktur, variationer i enzymstruktur och
reaktionsmediet på enantioselektiviteten för Candida antarctica lipas B studerats.
Bidrag från entalpi, ∆R-S∆H‡, och entropi, T∆R-S∆S‡, till den differentiella fria
aktiveringsenergin, ∆R-S∆G‡, för kinetiska resolveringar av sekundära alkoholer visades vara i
samma storleksordning. Generellt var de båda termerna motverkande, det vill säga att den
enantiomer som var gynnad av entalpi var missgynnad av entropi. 3-Hexanol var ett undantag,
där den enantiomer som enzymet föredrog var gynnad både entalpiskt och entropiskt.
Resolvering av 1-bromo-2-butanol visade att icke-steriska interaktioner påverkade både
∆R-S∆H‡ och ∆R-S∆S‡. Molekylmodellering av övergångstillståndet för enzym-substrat
komplexet visade på en korrelation mellan spatiell frihet och entropin för övergångstillståndet.
Acylkedjans längd påverkade enantioselektiviteten för sekundära alkoholer.
Punktmutationer i enzymets aktiva centrum visades kunna öka eller minska
enantioselektiviteten. Förändringarna orsakades av ändringar i både ∆R-S∆H‡ och ∆R-S∆S‡ som
delvis uppvägde varandra. Studier av varianter med enkel- och dubbelmutationer visade att de
observerade förändringarna inte var additiva.
Reaktionsmediet påverkade starkt enantioselektiviteten. Transförestring av en
sekundär alkohol katalyserad av Candida antarctica lipas B studerades i åtta vätskeformade
organiska lösningsmedel och superkritisk koldioxid. En korrelation mellan enantioselektivitet
och lösningsmedlets molekylära volym upptäcktes.
Differentiell aktiveringsentalpi, ∆R-S∆H‡, och entropi, ∆R-S∆S‡, påvisar ett
kompensativt förhållande. Denna kompensation är dock inte perfekt, vilket möjliggör
förändringar av enantioselektivitet. De termodynamiska parametrarnas komponenter är
mycket komplexa och beroende av varandra, men om deras roller klargörs kan rationell
design av enantioselektiva enzymatiska processer bli möjliga.
To my parents
List of papers
This thesis is based on the following publications, which in the text will be referred to by their
roman numerals.
I Overbeeke P. L. A., Ottosson J., Hult K., Jongejan J. A., and Duine J. A. The
temperature dependence of enzymatic kinetic resolutions reveals the relative
contribution of enthalpy and entropy to enzymatic enantioselectivity. Biocatal.
Biotransform. 17 (1999) 61-79
II Ottosson J., and Hult K. Influence of the acyl chain length on the enantioselectivity of
Candida antarctica lipase B and its thermodynamic components in kinetic resolution
of sec-alcohols. J. Mol. Catal. B: Enzym. 11 (2001) 1025-1028
III Ottosson J., Rotticci-Mulder J. C., Rotticci D., and Hult K. Rational design of
enantioselective enzymes requires considerations of entropy. Protein Sci. (2001)
In press
IV Ottosson J., Fransson L., and Hult K. Substrate entropy in enzyme enantioselectivity.
Submitted manuscript
V Ottosson J., Fransson L., King J. W., and Hult K. Size as a parameter for solvent
effects on Candida antarctica lipase B enantioselectivity. Submitted manuscript
Table of contents
INTRODUCTION
1
Background
Biocatalysis in non-aqueous media
1
2
Chirality
4
Enzyme enantioselectivity
7
Candida antarctica lipase B
11
Molecular modeling
14
ANALYSIS OF ENZYME ENANTIOSELECTIVITY
16
Theory
16
Role of the substrate
Thermodynamic analysis
Alcohol moiety
Molecular modeling of spatial freedom of the substrate
Acyl moiety
19
20
20
22
26
Role of the protein
Thermodynamic analysis
28
28
Role of the solvent
Thermodynamic analysis
Enthalpy-entropy compensation
31
32
36
Concluding remarks
38
ACKNOWLEDGEMENTS
39
REFERENCES
40
Introduction
The research in this thesis lies within the field of biocatalysis. Biocatalysis deals with
synthesis methods that exploit natural catalysts [1]. Catalysts are substances that increase the
rate of a reaction without themselves being consumed. Natural catalysts have been evolved by
nature since beginning of time to control and regulate almost all chemical reactions required
to sustain life. In biocatalysis they are put to work by and for mankind, often in conditions and
with substrates that are unnatural for them.
Background
Biocatalysis is by no means a modern concept. Brewing of beer, for example, predates
recorded history [1]. In ancient China and Japan biocatalysis was used for the production of
food and alcoholic drinks and in Europe the ancient Greek writer Homer described the
production of cheese through stirring milk with a twig from a fig tree. The twig leaked a
biocatalyst (ficin) that caused a coagulation of the milk [2]. It was however not until the
19th century that a more scientific approach to biocatalysis began. Berzelius defined catalysis
and acknowledged the action of biological catalysts (1830s) and Kühne termed these enzymes
(around 1870, from greek: in yeast). The unique specificity of enzymes was in 1894 suggested
by Fischer to be due to structural complementarity between the enzyme and its substrates. He
coined the “lock and key” mechanism with the analogy of the substrate fitting the enzyme like
a key to a lock [3, 4]. It was early recognized that enzymes do not require the environment of
the living cell [5]. The first reports of enzymatic reactions in organic solvents came already in
1898 and 1900 [6, 7]. However, it was not definitely settled that enzymes are proteins until the
first crystallizations of urease, pepsin, trypsin and chymotrypsin in the 1920s and 1930s [8, 9].
Although nearly all enzymes are proteins it is now known that RNA may display enzyme
activity (ribozymes) [4].
Nature has evolved enzymes to be very efficient and selective catalysts and it is
particularly these properties that make them so interesting for use in chemistry. Biocatalytic
processes are increasing in number [10] and include processes that utilize whole-cells as well
as purified enzymes. The rapid development and increased interest in the field is reflected by
and analyzed in the recent special issues of Nature and Chemical & Engineering News on
biocatalysis [1, 10-18]. Several industrial processes based on biocatalysis are in use and
1
although so far most of them are for low-volume, high-value products, such as pharmaceutical
synthons, there are exceptions such as the production of ethanol and fructose that are
produced annually at a million ton scale at a low price [16]. The production of aspartame, a
low-calorie sweetener, is another example of a large-scale biocatalytic process [10].
Biocatalysis in non-aqueous media
The use of enzymes as catalysts in organic solvents was for a long time met
with great skepticism among biochemists. This is not surprising as the addition of organic
solvents to aqueous media has been a technique to precipitate and denature proteins.
Consequently, the breakthrough of biocatalysis in non-aqueous media waited until the early
1980s [19-23]. Nowadays, it is well established that many enzymes, such as lipases, can
remain active and stable in pure organic solvents. Changing the hydrophobicity of a
predominantly aqueous media through addition of organic solvents cause the hydrophobic
effects, which keep the hydrophobic residues buried in the core of the protein, to rupture and
cause denaturation of the protein. When exposed to a pure organic solvent, many proteins do
not unfold but remain active. It has been suggested that proteins, folded in an aqueous
environment, are kinetically trapped in their native structure in organic solvents [13, 24-26].
In addition, in media of low water content, enzyme inactivation, caused by hydrolysis of
peptide bonds and deamidation of asparagine and glutamine residues, is reduced [13]. In fact
many times enzymes are more stable in organic solvents than in water. For instance, increased
thermostability in dry organic solvents with substantial increases in enzyme half-life at 100ºC
compared to water has been observed [22, 27]. There are a number of advantages of using
enzymes in organic solvents, Table 1. Many substrates that are insoluble in water can be
dissolved in organic solvents. That the enzyme itself is insoluble is often an advantage that
simplifies its recovery and reuse, many times an economically very important point. A change
from an aqueous environment can favor a shift in the equilibrium enabling synthetic reactions
to be achieved with hydrolytic enzymes. Moreover, the unique selectivity and activity of
enzymatic reactions is achieved under mild reaction conditions. This minimizes problems
with reaction conditions deleterious to the substrate or product, and avoids handling of harsh
chemicals or reaction conditions. The enzyme’s instability in harsher reaction conditions,
common in industry, can of course be a disadvantage. The search for enzymes from various
extremophilic organisms may, however, result in biocatalysts that can withstand more extreme
conditions [28, 29]. Furthermore, enzymes are environmentally benign and are unlike metal
2
Table 1. Benefits of biocatalysis in non-aqueous media
§
Increased substrate solubility
§
Simplified recovery of biocatalyst
§
Shift to synthetic reactions
§
Mild reaction conditions and minimization of side-reactions
§
Environmentally benign catalysts
§
High selectivity
§
Simplified work-up of products
§
Avoids microbial contamination
catalysts completely degraded in nature. Although some enzymes do display practically
perfect selectivity, many enzymes can accept a broad range of unnatural substrates still with
high chemo-, regio- or enantioselectivity. Additionally, the use of an organic solvent often
simplifies work-up procedures and avoids microbial contamination of the reaction medium
[13, 30].
Enzymes have also been shown to function in supercritical fluids, such as supercritical
carbon dioxide (SCCO2) [31, 32]. The use of SCCO2 as solvent for enzymatic reactions poses
a number of advantages. The technique is environmentally acceptable, leaving no toxic
solvent remains neither in product nor waste. Simple changes in pressure and temperature can
modulate the properties of the solvent, such as solubilizing capability and density.
Furthermore, workup after reaction can be greatly simplified as the technique is easily
combined with other unit operations such as supercritical extraction or fractionation. Mass
transfer limitations may be overcome in SCCO2 due to high diffusivity and low viscosity and
surface tension. A number of reviews on biocatalysis in supercritical media have been
presented [33-35].
Biocatalysis in non-aqueous media developed over the last two decades of the 20th
century into a well-established technique. Lipases are a class of enzymes that has proved to be
particularly successful biocatalysts. There are a number of commercial applications of lipases
both in aqueous media, such as detergent enzymes and in food and paper technology, as well
as in non-aqueous media, in the fine chemicals industry for production of various organic
intermediates [36, 37].
3
Chirality
An object that is not identical with its mirror image, although the two have identical
composition, is said to be chiral. This applies to macroscopic objects such as our hands but
also exists on the molecular level. Just like our left and right hands, both composed of a palm
and five fingers, are distinctly different, a molecule can be different to its mirror image.
Figure 1 illustrates how a molecule of 2-butanol in no way can be oriented to be identical to
its mirror image. The two molecules are said to be enantiomers.
C2H5
H
C
C2H5
C2 H5
OH
OH
CH3
C
H
CH3
CH3
C
C2 H5
OH
H
(S)-2-butanol
H
C
CH3
OH
Three orientations of (R)-2-butanol
mirror plane
Figure 1. The S and R enantiomers of 2-butanol are non-identical mirror images of
each other. (R)-2-butanol can not be rotated to be superimposable on (S)-2-butanol.
Enantiomers have identical physical and chemical properties and show the same
behavior in an achiral environment, but may exhibit very different behavior in a chiral one. As
chirality is a widespread phenomenon in nature there are many examples where enantiomers
function very differently, Figure 2. The sensitive receptors in the human nose will signal the
odor of lemons or oranges depending on which enantiomer of limonene they are exposed to.
Similarly, the enantiomers of carvone give the flavor of cumin or spearmint. In
pharmaceuticals often only one of the enantiomers of a chiral drug is the effective agent. The
other enantiomer may be less effective, totally ineffective, or in the worst case even toxic. The
world became terribly aware of the latter as a number of children were born with severe birth
defects in the late 1950s after pregnant mothers had been prescribed thalidomide for the
treatment of morning sickness. Thalidomide was administered as a racemic drug, that is it
consisted of equal amounts of both enantiomers. One of the enantiomers was the
pharmaceutical agent but the other caused developmental malformations. Later research has
indicated that the enantiomers actually are interconverted in the body [38]. Legislation now
4
O
*
OH
O
*
Ibuprofen
*
anti-inflammatory or ineffective
Carvone
*
flavor of cumin or spearmint
OH
Terpineol
O
Limonene
scent of lilac or cold pipe
scent of lemon or oranges
N
*
O
F
N
F
O
O
*
F
O
H
Thalidomide (Neurosedyn)
O
sedative or teratogenic
NH
O
Fluoxetine (Prozac)
*
antidepressant or ineffective
*
HS
OH
H
N
O
NH2
Penicillamin
antiarthritic or extremely toxic
Propranolol
OH
*
N
H
Ethambutol
b-blocker or cause sterility
H
N
O
*
HO
H
N
HO
tuberculosis drug or cause blindness
N
O
NH2
O
*
S
NH2
*
O
Aspargine
Sweet or bitter flavor
N
Omeprazole (Losec)
stomach ulcer drug or ineffective
O
Figure 2. Examples of chiral substances with distinctly different biological functions
of the enantiomers. The position of the stereocenter is marked with an asterisk.
requires both enantiomers of a racemate drug to be studied in detail for its pharmacological
effects
and
consequently,
the
market
for
enantiopure
chiral
drugs
is
rapidly
expanding [39, 40]. A number of formerly racemic drugs are or will soon be reintroduced as
single enantiomer drugs, for example Losec ((S)-omeprazole) and Prozac ((R)fluoxetine) [39]. These drugs can be marketed as free from the additional burden on
metabolism that the less effective enantiomer represents with the bonus of new patents. In the
light of the different effects of enantiomers in nature the importance of a nomenclature that
unmistakably distinguishes them from one another becomes evident. Cahn, Ingold and Prelog
5
developed such a nomenclature that designates one enantiomer as R and the other as S,
Figure 1 [41].
The similarities of enantiomers make them difficult to separate or produce in pure
form. Some of the classical methods are isolation from nature’s “chiral pool”, crystallization,
and chemical conversion to diastereomers, which are possible to separate with classical
techniques. However, the chiral pool of nature very often lacks the desired compound in high
enough quantities (if it exists at all) and few organic compounds form enantiopure
crystals [30]. Development of methods for asymmetric synthesis, that is synthesis of enantiopure or -enriched compounds, is consequently in high demand. For this biocatalytic methods
show great potential.
A measure of the degree of enantiopurity is the enantiomeric excess ee, Equation 1.
ee =
[R ] - [S ]
[R ] + [S ]
[R]: concentration of the R-enantiomer
(1)
[S]: concentration of the S-enantiomer
One way of analyzing ee is to measure the degree of rotation of plane polarized light, as
enantiomers rotate such light in opposite directions. However, to determine the enantiomeric
excess one needs to know the optical rotation at 100% ee, which is not the case when working
with a new substance. Chemical transformation of enantiomers with an enantiopure substance
allows preparation of diastereomers that are separable by ordinary methods and thus allows
determination of ee. Much of the troubles of analyzing enantiomers were overcome with the
advent of new chiral stationary phases for liquid and gas chromatography in the late 1980s
[42]. This has significantly simplified the analysis of enantiomers and consequently studies of
enantioselectivity.
Substances may also be prochiral, that is they are not chiral themselves but may form a
chiral product in a reaction. This is exemplified by 2-butanone (prochiral) in its reduction to
2-butanol (chiral), Figure 3. Depending on from which side of the carbonyl the reaction takes
place the R- or the S-enantiomer of the alcohol will form.
The high stereoselectivity of enzymes and the importance of chirality in many areas
make biocatalysis a rapidly growing technology for the production of enantiopure substances
[17, 43, 44]. Enantiopure secondary alcohols are important synthons for the preparation of
pharmaceuticals and agrochemicals for which several biocatalytic approaches have been
applied in industry [45]. Kinetic resolution of sec-alcohols using lipases often yields product
of high enantiopurity and is the reaction system studied in papers I-V.
6
OH
R
O 1
OH
2
S
Figure 3. Reduction of the prochiral substance 2-butanone yields either of the
enantiomers of 2-butanol.
Enzyme enantioselectivity
As enzymes themselves are chiral, they can act as the chiral environment needed to
preferentially catalyze one enantiomer of a chiral substrate or form a single enantiomer from a
prochiral substrate. Ogston developed a rationale for the enantioselectivity of enzymes, which
states that a substrate must be attached by at least three points on the enzymes for a high
enantioselectivity [30]. Figure 4 illustrates the preferred binding of a substrate that will yield
just one enantiomer of the product. In production of the other enantiomer the complementarity
to the enzyme is reduced. The enzyme has trouble accommodating the incorrect enantiomer
like a right hand does not fit a left hand glove. Consequently, the incorrect or non-preferred
enantiomer will require higher activation energy to react and will undergo catalysis at a lower
reaction rate, Figure 5. This allows for kinetic resolution, that is separation of enantiomers by
stopping the reaction before equilibrium is attained, yielding a mixture of enantio-enriched
reagent and product.
chiral substrate
prochiral substrate
prochiral substrate
R4
E1
R1
C
R1
R3
R2
E3
E1 R 1
R3
C
R2
E2
enantiomer selectivity
E3
E1
E2
enantioface selectivity
R1
C
R3
R2
E3
E2
enantiotopic selectivity
Figure 4. Three-point attachment rule of enzyme enantioselectivity. The
complementary interaction between substituents of the substrate (R) with enzyme
residues (E) enables catalysis yielding a single enantiomer from a chiral or prochiral
substrate.
7
G
∆GS
∆R-S ∆G
‡
‡
∆GR
‡
E+S
[TS]
‡
E+P
reaction coordinate
Figure 5. Schematic free energy profile of an enantioselective enzyme-catalyzed
reaction favoring the R-enantiomer.
The progress of the enantiomeric excess of substrate (eeS) and product (eeP) versus the
degree of conversion for a kinetic resolution is displayed in Figure 6. At low conversion the
enzyme experiences a nearly racemic concentration of the substrate (no or low eeS) and
practically only catalyzes the conversion of the preferred enantiomer yielding a high eeP. As
the reaction proceeds the concentration of the preferred enantiomer decreases and the enzyme
now experiences a relatively higher concentration of the other enantiomer, raising its
probability to be converted, eeS increases and eeP decreases. In order to have a parameter of
the enzyme that describes its degree of enantioselectivity, Chen et al. introduce the
enantiomeric ratio, E, and define it as the ratio between the specificity constants, kcat/KM, for
the enantiomers. E describes how well the enzyme discriminates between the enantiomers of a
substance under given reaction conditions. The shape of the progression curves of eeS and eeP
versus conversion thus changes with E, Figure 6. For the extreme values of E of highly
enantioselective enzymes the reaction proceeds to 50% conversion after which the reaction
rate drops so drastically that it practically stops. E can be determined from conversion, c, and
the enantiomeric excess of either the substrate or product, Equation 2 and 3 [46].
E=
ln[(1 - c )(1 - eeS )]
ln[(1 - c )(1 + eeS )]
(2)
E=
ln[1 - c(1 + eeP )]
ln[1 - c(1 - eeP )]
(3)
Estimations of E by single point measurements of eeS or eeP at one conversion are
many times unreliable as determination of conversion often lacks a high degree of accuracy.
8
Figure 6. Plot of enantiomeric excess of substrate, eeS, and product, eeP, as a
function of the conversion of an enzyme-catalyzed irreversible kinetic resolution
reaction. The plots were made with the computer program Selectivity developed by
the group of Professor Faber, University of Graz http://borgc185.kfunigraz.ac.at.
This may be overcome by the use of algorithms that calculate E by nonlinear regression of
data obtained at several extents of conversion [47, 48]. Determinations of conversion can be
avoided if E is calculated by another equation presented by Rakels et al., Equation 4 [49].
Equation 4 is a combination of equation 2 and 3 and was used in the present investigation.
æ
ç
1 - eeS
lnçç
ee
çç 1 + S
eeP
E= è
æ
ç
1 + eeS
lnçç
ee
çç 1 + S
eeP
è
ö
÷
÷
÷
÷÷
ø
ö
÷
÷
÷
÷÷
ø
(4)
These, and other methods for the evaluation of E and their advantages and disadvantages, are
reviewed by Straathof and Jongejan [50]. Equation 2-4 and Figure 6 applies to irreversible
reactions. Theoretical treatment of reversible reactions, which incorporates the equilibrium
constant, Keq, to the equations, has also been presented [51].
9
The relation between the difference in Gibbs free energy of the transition states of the
enantiomers, Figure 5, and the enantiomeric ratio E can be shown with transition state theory
to be
D R -S DG ‡ = DGR‡ - DGS‡ = -RT ln E
(5)
Enzyme enantioselectivity provides an excellent opportunity for fundamental studies of
enzyme catalysis due to the identical ground state energies of enantiomers, which reduces the
comparison of their reaction path energetics to the difference of the two enzyme-substrate
transition states. Measurements of absolute rates are often problematic for instance under
denaturing conditions where the amount of active enzyme is difficult to determine. In
enantioselective systems determination of relative reaction rates are sufficient.
It is important that the determination of E is reliable when comparing and studying the
performance of enzymes in enantioselective catalysis. Any conclusions on enzyme
mechanism and performance drawn from changes in E may otherwise be faulty. Although the
comparison of the enantiomeric excess of reaction products is a very good measure of product
quality it can not be used to make any conclusions about enzyme enantioselectivity unless the
enantiomeric excess are determined at identical degrees of conversion. In studies on enzyme
enantioselectivity, the enantiomeric ratio, E, should be the parameter compared. Particular
precautions should be made when determining high E-values, as these are difficult to measure
with a high degree of accuracy because analysis of very high enantiomeric excess values are
required.
Enzyme enantioselectivity is apart from the reaction conditions influenced by the
substrate structure, variations on the catalyst itself and by the reaction medium. Consequently,
much effort has been made to optimize enantioselectivity with substrate, protein and solvent
engineering. For lipases, Berglund recently presented an overview of means to control
enantioselectivity [52]. In this thesis some of the parameters that affect the performance of
enantioselective enzymes have been investigated using thermodynamic analysis of observed
changes in enantioselectivity. The ultimate goal has been to provide additional information for
a molecular understanding of enzyme enantioselectivity in particular and enzyme catalysis in
general.
10
Figure 7. Structure of Candida antarctica lipase B. The catalytic serine is displayed
in ball-and-stick representation.
Candida antarctica lipase B
With the aim of finding enzymes with extreme properties, the yeast Candida
antarctica was isolated from a sample from Antarctica [53]. This organism produces two
lipases (E.C. 3.1.1.3 triacyl glycerol hydrolases) with designations A and B. In 1994, the
structure of CALB (Candida antarctica lipase B) was determined by X-ray crystallography,
Figure 7 [54]. It is composed of 317 amino acid residues and has a molecular weight of
33 kDa, which makes it a fairly small protein compared to other lipases. Like all other lipases
of known structure it belongs to the α/β-hydrolase fold family of proteins [36, 55]. Contrary to
most other lipases this enzyme does not display interfacial activation, that is an increase in
activity caused by exposure to a water-lipid interphase [36, 56]. This is considered to be due
to the fact that this enzyme lacks the lid that regulates the access to the active site that is a
common feature in other lipases. Another unusual feature of this lipase is that it does not have
the consensus amino acid sequence, GXSXG, around the active site serine that is common to
11
O
Asp187
O
Ser105
H
N
N
O
R
O
O
Asp187
O
R1
Acylation
Free enzyme
R1
O
H
H
His224
Ser105
O
N
N
O
H
O
Oxyanion
hole
R
His224
Tetrahedral intermediate
O
R2
R1
O
R-OH
Asp187
Asp187
Deacylation
O
O
Ser105
R1
O
O
H
N
His224
N
O
H
O
Ser105
O
O
Oxyanion
hole
H
N
N
R1
O
R2-OH
R2
His224
Tetrahedral intermediate
Acyl enzyme
Figure 8. Catalytic mechanism of a Candida antarctica lipase B catalyzed
hydrolysis (R2=hydrogen) or transesterification (R2= alkyl).
this folding family. Instead it has a threonine at the position of the first conserved
glycine [54]. Similar to other serine hydrolases a serine-histidine-aspartate catalytic triad is
responsible for the catalytic action with the mechanism outlined in Figure 8. It is a two-step
mechanism with an acylation step and a deacylation step separated by a covalent acyl-enzyme
intermediate. CALB has proven to be a particularly useful biocatalyst of good stability and
activity that catalyze a diverse range of reactions. It has been shown to exhibit high regio- and
enantioselectivity. Anderson et al. published a review article in 1998 on the use of this
enzyme in organic synthesis [53]. Papers I-V present a study of the kinetic resolution of
sec-alcohols catalyzed by CALB. The enantiopreference of the enzyme follows Kazlauskas’
rule, Figure 9 [57]. An overview of the preparation of enantioenriched sec-alcohols using
CALB, including protocols, has previously been presented by us [58].
H
M
OH
L
Figure 9. The preferred enantiomer of a sec-alcohol for hydrolases that follow the
rule of Kazlauskas [57]. L= large sized substituent, M= medium sized substituent.
12
Protein surface
Stereospecificity
pocket
L
O
HN
O
N
H
O
H
H
O
H
O-Thr40
N-Thr40
Asp187
O
HN
O
N
H
O
H
H
O
Ser105
O
N-Gln106
Acyl site
H
L
Ser105
O
M
Stereospecificity
pocket
H
M
Asp187
Alcohol site
Acyl site
Alcohol site
Protein surface
H
O-Thr40
N-Thr40
N-Gln106
His224
His224
Figure 10. Overview of the two CALB binding modes for the enantiomers of secalcohols in the tetrahedral intermediate transition states presented by Orrenius et al.
[60]. The fast reacting enantiomer in the left panel is less sterically hindered. The
hydrogen bonding network in the active site is required for catalysis. At least two of
the three potential hydrogen bonds to the backbone of Thr40 and Gln106 and the side
chain of Thr40 in the oxyanion hole are considered necessary [54].
The crystal structures of CALB with Tween 80 and in a covalent complex with a
phosphonate inhibitor, in combination with molecular modeling studies have revealed the
existence of a stereospecificity pocket within the active site [59]. This pocket, located deep in
the active site has been suggested to harbor one of the substituents on the stereocenter of
sec-alcohols. Further modeling and experimental studies brought forward a model of two
different productive binding modes for the enantiomers in transition state [60]. These different
modes are illustrated in Figure 10. According to the model, the two modes are necessary to
allow both enantiomers to develop the hydrogen-bonding pattern within the active site,
required for catalysis [54, 60]. The fast reacting enantiomer, generally R, places its mediumsized substituent of the alcohol moiety into this pocket and its large substituent in the volume
of the active site open to the surface of the protein. In order to keep the above-mentioned
hydrogen-bonding pattern, the slow reacting enantiomer, usually S, has to place its large
substituent into the stereospecificity pocket instead. A steric difference between the transition
states of the enantiomers can thus explain the enzyme’s ability to resolve enantiomers.
However, this model is static and only considers differences in activation enthalpy and not
entropy. Thermodynamic analysis of enantioselectivity has revealed that entropy as well is an
important parameter for the success of the discrimination between enantiomers exhibited by
an enzyme [61].
13
Molecular modeling
The fast development of computational chemistry has been of great importance for
studies on proteins. The graphical tools alone, that enable visualization of proteins whose
structures have been determined, have had a great impact on our understanding of proteins.
The number of proteins in the structural databank [62] is rapidly increasing allowing more and
more proteins to be analyzed and studied with a range of methods that computational
chemistry provides.
Quantum chemistry calculations on whole proteins require more computational power
than is feasible today. This may be overcome by the use of empirical force field methods,
based on classical mechanics and often termed molecular mechanics. The potential energy of
a molecule is calculated as a sum of interactions related to bond stretching, torsions and angle
bending as well as non-bonded interactions such as electrostatic and van der Waals
interactions [63]. The force field parameters are based on experimental data and a number of
different force fields suitable for different types of compounds have been developed [63]. As
electrons are not treated explicitly with molecular mechanics, chemical reactions in which
bonds are formed and broken cannot be studied. Some combinational approaches where the
reactive parts of the molecule, such as the catalytic residues of an enzyme, are treated with
quantum mechanics and the rest of the molecule with molecular mechanics have been
successful [63].
Molecular dynamics (MD) simulation calculates the time-dependent dynamics of a
molecule. In the simulations the movements of the atoms in a force field are calculated with
Newtonian physics in very short time steps. As the time-steps have to be in the order of
femtoseconds, the real time that may be simulated with reasonable computer capacity is
limited by the number of atoms in the system and the computational force at hand. Another
method for exploring possible conformations is to perform a systematic search by rotating
bonds in increments and determine the force field energy of each conformation. This method
is independent of any energetic barriers between the conformations, but is limited by
computational power as the number of calculations will increase exponentially with the
number of bonds included. Consequently, the method is limited for analysis on the
conformations of only a few atoms.
Molecular modeling can help to explain on a molecular level enzyme selectivities and
can suggest changes for improvement of biocatalyst performance. In the long run a major goal
is to be able to correctly quantify the degree of selectivity, which could save time and
14
resources today spent on empirical screening. Enantioselective enzymes are especially well
suited for molecular modeling studies as the identical ground state energies of enantiomers
allow comparison of only transition state structures and energies. A review by Kazlauskas on
molecular modeling on biocatalysts was published in 2000 [64].
15
Analysis of enzyme enantioselectivity
Presently, when designing a biocatalytic process, much time and resources are spent
on optimization and screening. Therefore, knowledge that would enable predictions on the
outcome of the reactions would be most beneficial. A lot of such information and knowledge
has been acquired since the breakthrough of biocatalysis in the 1980s and several empirical
and modeling based rules and criteria are now available [30, 65, 66]. However, much is still
not predictable and our understanding at a molecular level of the catalytic events and the
factors influencing them are still incomplete. Enantioselective systems are, as previously
mentioned, ideal for fundamental studies of enzyme action due to the identical ground state
energies of the competing substrates (the enantiomers). Thermodynamic analyses, through
studies of the variation of enantioselectivity with temperature, provide additional information
on the role of activation enthalpy and entropy that hopefully will yield an increased molecular
understanding.
Theory
Enantioselectivity is caused by a difference in activation free energy between the
enantiomers that may be separated into an enthalpic and an entropic term, Equation 6,
Figure 11.
D R -S DG ‡ = D R -S DH ‡ - TD R -S DS ‡ = -RT ln E
(6)
Originally the contribution of entropy was considered negligible as both enantiomers are
significantly restricted in the small volume of the enzyme active site [67-69]. DeTar estimated
in 1981 the maximal entropy contribution to enantioselectivity for a hydrolysis reaction
catalyzed by chymotrypsin [67]. Assuming the available volume of the active site to be no
more than twice the volume of the substrate he estimated the maximal contribution of
translational entropy to be Rln2 (giving a TDR-SDS‡ of 1.7 kJ mol-1 at 300 K). Contributions of
rotational entropy were excluded in the estimation with the argument that no rotational
freedom exists for complexed substrates. Differences in the contribution to the differential
free energy from vibrational entropy, TDR-SDS‡, were estimated by DeTar to be less than
4.2 kJ mol-1. In 1992 Phillips pointed out that DR-SDS‡ in fact can contribute significantly to
enantioselectivity [61].
16
G
S
H
∆HS
‡
∆G
∆R-S ∆G
‡
S
‡
∆R-S ∆H
‡
∆HR
∆S S
‡
∆G
‡
R
‡
∆S R
‡
E+S
[TS]
‡
E+P
E+S
reaction coordinate
‡
[TS]
E+P
E+S
reaction coordinate
∆R-S ∆S
‡
[TS]
‡
E+P
reaction coordinate
Figure 11. Reaction profiles for an enzyme catalyzed reaction of a chiral substrate
with enantiomers designated R and S. G = Gibbs free energy, H = enthalpy and
S = entropy.
The difference in activation enthalpy and entropy between the enantiomers, Figure 11,
can be determined by studying the variation of the enzyme’s enantiomeric ratio, E, with
temperature. Equation 6 rearranges to Equation 7
ln E = -
D R -S DH ‡ 1 D R -S DS ‡
× +
R
T
R
(7)
The enantiomeric ratio (or rather lnE) will vary with reciprocal temperature to an extent
determined by the enthalpic term (the slope of Equation 7, DR-SDH‡/R) at a level determined
by the entropic term (the intercept of Equation 7, DR-SDS‡/R). Differential activation enthalpy
is related to differences in the complementarity of each enantiomer in the transition state and
comprises steric and electrostatic interactions between the reaction components, the enzyme,
its substrate and the solvent. Differential activation entropy, however, is more difficult to
visualize and comprises differences between the enantiomers in conformational degrees of
freedom of the protein, losses in conformational entropy of the substrates or solvation. Figure
12 presents an overview of the theoretically possible relations of lnE or DR-SDG‡ with
temperature according to equations 6 and 7. The differential activation parameters can be
either positive or negative depending on which enantiomer is favored by enthalpy and
entropy. A temperature at which there is no enantioselectivity, E=1 and DR-SDG‡=0, the
racemic temperature, Tr, is determined as the ratio of the differential activation enthalpy and
entropy, Equation 8.
Tr =
D R -S DH ‡
D R -S DS ‡
(8)
17
‡
‡
‡
‡
∆∆H <0 ∆∆S >0
∆∆G
‡
lnE
‡
‡
∆∆H <0 ∆∆S <0
Tr
1/Tr
T
0
‡
∆∆H <0 ∆∆S <0
1/T
0
‡
∆∆H <0 ∆∆S >0
Figure 12. Schematic overview of the theoretically possible relations between
differential activation free energy, ∆∆G‡, and temperature (Equation 6) or lnE and
reciprocal temperature (Equation 7). In the opposite enantiopreference both ∆∆H‡
and ∆∆S‡ change sign.
Generally, the enthalpic and the entropic contributions counteract each other, that is,
they are either both positive or both negative. This is in line with the notion that a better
transition state binding, low enthalpy, comes with a more rigid transition state, low entropy. In
other words the enantiomer favored by enthalpy is disfavored by entropy, Figure 11. Below
Tr, the major term in the differential free energy is the enthalpic term, Equation 6. The enzyme
will preferentially catalyze the enantiomer favored by enthalpy, and enantioselectivity will
decrease with temperature as the entropic term (TDR-SDS‡) of Equation 6 increases. Above Tr,
however, the major term is the entropic and the enzyme’s enantiopreference is changed to the
enantiomer favored by entropy and enantioselectivity increases with temperature as TDR-SDS‡
increase. Theoretically, the differential activation enthalpy and entropy may be of different
signs, that is the favored enantiomer is both favored by enthalpy and entropy. This situation
produces a negative Tr with no physical meaning.
Several studies of the temperature dependence of enzyme enantioselectivity are now
available in literature [61, 70-77] and it has become evident that the entropic term is by no
means negligible, but actually has a strong influence on enantioselectivity. Generally enzyme
catalyzed reactions have racemic temperatures above the experimental temperature, and
enantioselectivity decreases with increasing temperature. However, several examples of
reactions above the racemic temperature with the somewhat counterintuitive property of
increasing
enantioselectivity
with
increasing
reaction
temperature
have
been
presented [71, 74, 78-84]. These systems are interesting as one may increase both
enantioselectivity as well as activity by simply increasing the reaction temperature. Some
18
reaction systems with racemic temperatures within the experimental temperature range have
also been identified [73, 84]. By a change in reaction temperature the enzyme’s
enantiopreference could be switched to the other enantiomer. The situation in which the
differential activation parameters have opposite signs and favor the same enantiomer seems to
be rare but a few such cases are found in the present investigation (papers I and IV) and
elsewhere [76, 85].
The observed variations in the components of enantioselectivity, ∆R-S∆H‡ and ∆R-S∆S‡,
are related to shifts in the individual activation parameters, ∆H‡ and ∆S‡, for one or both
enantiomers. Determinations of individual activation parameters are possible but require pure
enantiomers and determination of absolute reaction rates, an experimentally more complex
task.
Enzymatic reaction systems can be said to consist of three components; the substrates,
the protein and the solvent. These three components interact and it is their overall interactions
that determine the success of the catalysis. The effects on enantioselectivity and the
differential activation parameters from each of these components are complex and highly
interdependent. However, for the sake of discussion an attempt to treat each component
separately is made in the following chapters.
Role of the substrate
Much effort has been put into setting up rules and criteria, based on the substrate
structure, to predict enzyme selectivity. The rule of Kazlauskas for prediction of
enantiopreference, Figure 9, [57] and various boxmodels [86-89] based on experimental data
from many substrates with varying structure are successful examples. Moreover, several
molecular
modeling
studies
have
investigated
the
structural
requirements
of
substrates [60, 69, 90-93]. However, the influence of entropy is not considered in these
approaches. The chiral environment of the enzyme active site will affect enantiomers
differently. Differences in steric and electrostatic interactions with the enzyme will give rise
to differential activation enthalpy, whereas differences in the degrees of freedom of the
activated enzyme-substrate complexes will yield a differential activation entropy.
Understanding the structural requirements of substrates for good enantioselectivity is of great
value and is used in substrate engineering for improved enantioselectivity. For example,
enantioselectivity can be improved by the use of suitable protective groups [30] or forming
salts [94].
19
Thermodynamic analysis
In one of the early works that included a thermodynamic analysis in the study of
parameters
important
to
enantioselectivity,
Zheng
and
Phillips
report
enhanced
enantioselectivity of a dehydrogenase with coenzyme analogs [95]. Analogs to the native
coenzyme NADPH yield in reductions of 2-butanone catalyzed by alcohol dehydrogenase
from Thermoanaerobacter ethanolicus, SADH, higher enantioselectivity. Kinetic experiments
of the reverse reaction and determination of the temperature dependence of E reveal that the
observed increase in enantioselectivity, when changing cofactor, is caused by changes in both
DR-SDH‡ and DR-SDS‡. In another study of the same enzyme, the change of substrate from
2-butanol to 2-pentanol, reveal changes in both the enthalpic and the entropic term [70]. For
2-butanol the racemic temperature is 26°C and consequently, the stereochemical outcome of
the reaction can be set by the choice of reaction temperature. Below 26°C the S-isomer is the
preferred enantiomer and above the R-isomer. For 2-pentanol the racemic temperature is 77°C
and both DR-SDH‡ and DR-SDS‡ are lower compared to those of 2-butanol. The change of
coenzyme and the addition of a methyl group on the substrate both cause changes in the chiral
alcohols’ interactions with the enzyme or solvent as well as the degrees of freedom of the
enzyme-substrate complex.
Kinetic resolutions of sec-alcohols with transesterification reactions catalyzed by
CALB involve two substrates, an acyl donor and the chiral alcohol as the acyl acceptor,
Figure 8. Both substrates affect enantioselectivity, as the acyl chain and the alcohol both are
part of the chiral discrimination transition state. Much is known of the structural requirements
of the alcohol for high enantioselectivity for this enzyme, which has led to the model of
Orrenius et al. presented on page 13 [60]. Entropy factors are not considered in this model,
but it provides a molecular level picture of the enzyme-substrate transition states, most
valuable in studies of the enzyme. The effects of substrate structure on CALB
enantioselectivity were studied with thermodynamic analyses in papers I and IV (variations
on the alcohol moiety) and in paper II (variations on the acyl moiety).
Alcohol moiety
Table 2 presents E and its thermodynamic components for various sec-alcohols
(papers I and IV). For all the substrates a significant contribution of the differential activation
entropy to enantioselectivity was observed. In fact, the entropic term, TDR-SDS‡, was 25-61%
of the enthalpic term, DR-SDH‡, in the differential free energy of activation, DR-SDG‡. Small
20
Table 2. Enantiomeric ratio, E, and its thermodynamic components for the transesterification
of sec-alcohols catalyzed by Candida antarctica lipase B (paper I and IV). Vinyl octanoate
was the acyl donor for all but 1-bromo-2-butanol, where vinyl propionate was used.
Alcohol moiety
E
∆R-S∆G‡
∆R-S∆H‡
T∆R-S∆S‡
∆R-S∆S‡
Tr
298 K
298 K
298 K
kJ/mol
kJ/mol
kJ/mol
J/mol,K
K
1-bromo-2-butanol
100
-11.4
-23.5 ± 1.0 -12.1 ± 1.0 -40.7 ± 3.2 580
3-hexanol
340
-14.4
-9.0 ± 0.7 +5.5 ± 0.7 +18.3 ± 2.4 -490
2-butanol
8.3
-5.3
-10.7 ± 1.4
-5.4 ± 1.4 -18.2 ± 4.6 590
3-methyl-2-butanol
810
-16.6
-24.3 ± 1.1
-7.7 ± 1.0 -26.0 ± 3.4 940
3,3-dimethyl-2-butanol
460
-15.2
-20.4 ± 1.3
-5.2 ± 1.3 -17.6 ± 4.2 1160
Errors were calculated from the standard errors of the linear regression of Equation 7 and E was calculated from
the linear relation.
changes in the alcohol structure, such as an additional methyl or bromo group, brought about
changes in both DR-SDH‡ and DR-SDS‡, resulting in a wide range of E. The racemic
temperature, Tr, was in all cases above the experimental temperature except for 3-hexanol,
which had a negative Tr. 3-Hexanol displayed as the other substrates a negative differential
activation enthalpy but a positive differential activation entropy. In other words the preferred
enantiomer of this substrate (R) was both favored by enthalpy and by entropy. This is a
somewhat unexpected observation as one generally associates a low activation enthalpy
barrier with a low activation entropy barrier. For all the other alcohols in Table 2 the enthalpic
and the entropic term counteracted each other, that is, the enantiomer favored by enthalpy (R)
was disfavored by entropy. 1-Bromo-2-butanol has two isosteric substituents (-CH2CH3 and
-CH2Br) at the stereocenter but CALB still displayed a significant enantioselectivity to this
compound [96]. Consequently, non-steric interactions in the active site contributed to the
enantiodiscrimination and did so with both a differential activation enthalpy and
entropy, Table 2. Repulsive forces between enzyme residues within the stereospecificity
pocket and the halogen atom as it points into the pocket have been proposed to be the basis of
the observed enantiodiscrimination [96, 97]. Another example showing that non-steric
interactions contribute to enantioselectivity was recently presented for Candida rugosa lipase
MY [78]. In that study halogen substituents rather remote of the chiral center are shown to
affect the resolution of a chiral acid compared to their isosteric counterparts. It is difficult to
visualize the cause of the differential activation entropy for 1-bromo-2-butanol as the “large
and medium sized” substituents are isosteric. The potential spatial freedom in the active site is
the same for both enantiomers, independent of the two modes of binding, which suggests
similar transition state entropy. This is not the case, however, instead this substrate displayed
the largest entropic component of E in this series of alcohols. Non-steric interactions between
21
the halogen and the enzyme or the solvent may restrict the degrees of freedom experienced by
one enantiomer more than the other. Unfortunately, the lack of parameters in the force field
describing bromo-atoms excluded this substrate from the molecular modeling study of the
substrate accessible volumes in transition state described in paper IV.
Molecular modeling of spatial freedom of the substrate
The spatial freedom of a substrate in the active site may be estimated with various
molecular modeling methods. For estimation of differential activation parameters in enzyme
enantioselectivity it is sufficient to model only transition states, as enantiomers have identical
ground state energies. In paper IV the correlation between differential spatial freedom of the
enantiomers in the activated enzyme-substrate complex with the differential activation
entropy was investigated. Two different molecular modeling approaches were used to verify
whether the largest substrate accessible volume of a pair of enantiomers could be correlated to
the one with the highest activation entropy. Unfortunately, lack of parameters describing
bromo atoms in the force field prevented studies on 1-bromo-2-butanol. This was particularly
unfortunate as it would be most rewarding to study an isosteric compound with this approach.
One method to sample possible conformations of the substrate in the active site is
molecular dynamics, MD, simulations of the tetrahedral intermediate structure as a model of
the transition state, Figure 8. By MD simulations distinctly different conformations of
sec-alcohols in the active site of CALB could be identified (paper IV). The large substituents
pointed into different regions of the active site during the MD simulations for both
enantiomers of 2-butanol as well as (R)-3-hexanol and (R)-3-methyl-2-butanol, see Figure 4 in
paper IV. Some atoms of the different substrates occupied similar or identical positions in the
active site. For example, the two possible positions of carbon number three, the γ-carbon, on
the ethyl substituent of (S)-2-butanol were identical with the positions of the two γ-carbons of
the iso-propyl substituent of (S)-3-methyl-2-butanol. Similarly, the γ-carbon of the propyl
substituent of (R)-3-hexanol occupied both positions that were available for (R)-2-butanol. For
(S)-3-hexanol, however, only one position of the γ-carbon was identified. Possibly, one
conformation was prevented due to steric hinders of placing the propyl substituent into the
stereospecificity pocket.
MD simulations of 500 ps were not enough to reach a steady value for the
accumulated volume that the substrate had occupied during the simulation. This indicated that
all of the conformational space had not been sampled within the time frame of the simulation
in spite of that simulations of 500 ps are rather long for contemporary MD-studies [91-93].
22
900
819
800
795
721
737
775
729
715
700
684
Volume (Å3)
600
500
400
300
200
100
0
3-hexanol
2-butanol
R
R
S
S
3-methyl-2-butanol
R
S
3,3-dimethyl-2butanol
R
S
Figure 13. Substrate accessible volume in the active site of CALB determined with
500 ps molecular dynamics simulations of the octanoate esters of the sec-alcohols
(paper IV). The grey bars represent the enantiomer with highest transition state
entropy, see Table 2.
Nevertheless, the largest substrate accessible volume corresponded to the enantiomer with the
highest activation entropy for three of the four substrates, Figure 13. For 3,3-dimethyl2-butanol the substrate accessible volume did not correlate to the activation entropy. This
bulky substrate is very restricted in the active site and the tert-butyl group did not rotate
during the MD-simulation, Figure 14a. This was true for both enantiomers in their respective
binding modes. Even some methyl groups displayed restricted rotation, Figure 14b. Possibly,
the degree of insufficient sampling of the conformational space was more pronounced for this
substrate. This substrate had a reaction rate several hundred times lower than the other
substrates. One may expect the required simulation times to be correspondingly much longer
for a comparable estimation of the substrate accessible volume of the active site.
A second methodology to estimate the substrate accessible volumes and correlate it to
DR-SDS‡ was investigated in paper IV. Through systematic searches by rotation of the
substrate bonds an evaluation of the available volume was achieved. This method had the
advantage of having no relation to the time frame of the actual physical events as in MD
simulations. One drawback of this method was that the number of bonds and degrees of the
rotation increments must be restricted for a computationally feasible problem. Hence only
23
a)
c)
b)
Figure 14. (S)-3,3-dimethyl-2-butanol was highly restricted in the active site of
Candida antarctica lipase B during a 500 ps molecular dynamics simulation. a) The
tert-butyl substituent did not rotate. Hydrogens are not displayed. In panel b) and c)
two different methyl groups of this substrate are magnified, which display restricted
and free rotation respectively. Carbon in blue and hydrogens in green, yellow and
red. (paper IV)
a)
c)
b)
Figure 15. Estimations of the substrate accessible volume of (S)-2-butanol. a) A
single systematic search b) multiple (five) systematic searches and c) 500 ps
molecular dynamics simulation. Hydrogens are not displayed, oxygen atom in black,
carbon atoms in yellow red and blue (paper IV).
rotations of bonds within the substrate were studied with the consequence that the enzyme
was static within each systematic search. This is a very crude assumption as the enzyme is
likely to move to accommodate the substrate in the transition state. To overcome this
disadvantage several systematic searches were performed using different starting
conformations of the enzyme-substrate structure. A more complete estimation of the total
substrate accessible volume was achieved if the searches were overlapped. This is clearly
visible in Figure 15a and 15b. The largest substrate accessible volume was found with this
method for the entropically favored enantiomer of 2-butanol and 3-hexanol, Table 3. In this
study, the starting conformations for the systematic searches were sampled structures from
24
Table 3. Volume (Å3) within the active site explored by the alcohol part of the substrate in
systematic searches (paper IV). The total volume was calculated by superimposing the
structures from all individual systematic searches. A comparison with the substrate accessible
volume determined with a molecular dynamics (MD) simulation is shown in the last row.
Search
Volume (Å3)
Substrate
3-hexanol
2-butanol
3-methyl3,3-dimethyl2-butanol
2-butanol
R
S
R
S
R
S
R
S
1
290
284
138
149
178
170
186
180
2
282
293
142
164
180
168
199
177
3
259
274
164
179
181
167
196
174
4
283
295
147
167
170
173
188
191
5
298
287
145
156
173
171
206
180
6
294
279
7
288
271
8
289
306
9
320
274
10
292
298
379
342
174
188
188
185
210
194
Total
MD (V50ps)
298
253
182
194
189
173
202
199
V50ps represents the volume explored between time 50 and 500 ps in the MD simulation
MD simulations. The lack of correlation for 3-methyl-2-butanol and 3,3-dimethyl-2-butanol
may possibly be overcome by increasing the number of systematic searches and using more
widely different starting conformations than the MD simulation generates. With systematic
searches, energy barriers between different conformations are readily crossed and
consequently a more complete sampling of the substrate accessible volume of the active site
may be achieved. Comparing the resulting substrate-accessible volume achieved by the two
methodologies, one notes that although multiple systematic searches are required for a
complete analysis, a faster and more thorough analysis is achieved by this method, Figure 15b
and 15c.
It is practically impossible to separate the contribution to DR-SDS‡ related to restrictions
of the substrate from other causes such as differential solvation effects or differential
restriction of protein residues. Possibly, molecular modeling approaches that also take these
effects into account could solve this separation in the future.
25
Table 4. Enantiomeric ratio, E, and its thermodynamic components for transesterifications of
3-methyl-2-butanol with different vinyl esters catalyzed by Candida antarctica lipase B
(paper II).
Acyl moiety
E
∆R-S∆G‡
∆R-S∆H‡
T∆R-S∆S‡
∆R-S∆S‡
Tr
298 K
298 K
298 K
kJ/mol
kJ/mol
kJ/mol
J/mol,K
K
vinyl propionate
470
-15.3
1620
-18.7 ± 1.9
-3.4 ± 1.9 -11.5 ± 6.1
vinyl butanoate
390
-14.8
800
-23.6 ± 1.2
-8.2 ± 1.2 -29.6 ± 4.0
vinyl hexanoate
720
-16.3
1110
-22.3 ± 1.4
-6.0 ± 1.4 -20.1 ± 4.5
vinyl octanoate
810
-16.6
940
-24.3 ± 1.1
-7.7 ± 1.0 -25.9 ± 3.4
vinyl isobutanoate
480
-15.3
1060
-21.2 ± 1.8
-6.0 ± 1.7 -20.0 ± 5.7
octanoic acid
2500
-19.4
700
-33.7 ± 3.6 -14.3 ± 3.5 -48.1 ± 11.6
Errors were calculated from the standard errors of the linear regression of Equation 7 and E was calculated from
the linear relation.
Acyl moiety
In the narrow active site of CALB the acyl and alcohol moieties bind in a
hairpin orientation, Figure 10. It is thus not surprising that the acyl chain of the substrate
interacts with the alcohol part of the substrate and consequently influences the chiral
discrimination of alcohols. Hæffner et al. showed in molecular modeling calculations that
even remote amino acid residues in the acyl binding site positioned far from the stereocenter
interact differently with the enantiomeric transition states and thus contribute to
enantioselectivity [98]. Raza et al. studied the influence of acyl chain length on
enantioselectivity with molecular modeling and observed a significant difference in potential
energies between the enantiomers as a function of the length of the acyl chain [93]. The acyl
chain length of vinyl esters employed as acyl donors was shown experimentally to have a
strong influence on E and its thermodynamic components (paper II). The highest
enantioselectivity was achieved with the longest acyl chain, vinyl octanoate. E then decreased
with chain length for the hexanoate and the butanoate but increased again somewhat for the
vinyl propanoate, Table 4. The dissection of enantioselectivity into its thermodynamic
components revealed that their individual contributions were not straightforward. The longer
acyl chains (butanoyl, hexanoyl and octanoyl) differed mainly in their entropic components,
but vinyl propanoate displayed also a significant change in the differential activation enthalpy.
The resolution with a branched acyl donor, as in vinyl isobutanoate, yielded similar
enantioselectivity as vinyl propanoate but with activation parameters of values intermediate
between vinyl propanoate and vinyl butanoate (unpublished data). The molecular modeling
study in paper IV revealed that the spatial freedom of the acyl chain (octanoate) in transition
26
700
614
600
573
571
558
566
559
528
500
Volume (Å3)
500
400
300
200
100
0
3-hexanol
R
S
2-butanol
R
S
3-metyl-2-butanol
R
S
3,3-dimetyl-2-butanol
R
S
Figure 16. Substrate accessible volume for the acyl part of the tetrahedral
intermediate in the 500ps molecular dynamics simulations of octanoate esters (paper
IV).
state was different for the enantiomers as well as for different alcohol moieties, Figure 16.
Kinetic resolution of 3-methyl-2-butanol with octanoic acid as acyl donor showed surprisingly
high enantioselectivity in comparison with vinyl octanoate (unpublished data). This increased
enantioselectivity was caused by a significant increase in the absolute value of differential
activation enthalpy that was accompanied by a significant increase in the absolute value of the
counteracting differential activation entropy. However, DR-SDS‡ did not increase to the same
extent in order to completely cancel the effects of the increase in DR-SDH‡. This observed
difference in E and its differential activation parameters between the reaction with octanoic
acid and vinyl octanoate is very intriguing as the chiral discriminatory transition states of
these two reactions are identical, Figure 8. The acid may have affected the apparent pH of the
aqueous microenvironment around the active site, or interacted with any ionized groups of the
protein or in some other way affected the structure or flexibility of the enzyme [99].
27
Table 5. Enantiomeric ratio, E, and its thermodynamic components for the kinetic resolution
of 3-methyl-2-butanol catalyzed by Candida antarctica lipase B and variants with point
mutations (paper III). Vinyl octanoate was used as the acyl donor. ∆∆∆H‡ and ∆∆∆S‡
represents the change in ∆R-S∆H‡ and ∆R-S∆S‡ relative to wild-type.
E
∆R-S∆G‡
∆∆∆H‡
T∆∆∆S‡
∆R-S∆S‡
∆R-S∆H‡
Tr
Enzyme
296 K
Wild type
T103G
W104H
T42V
S47A
T42V/S47A
1000
2100
150
500
340
390
296 K
kJ/mol
-16.9
-18.9
-12.3
-15.4
-14.3
-14.7
kJ/mol
-20.8 ± 0.9
-31.7 ± 0.1
-44.7 ± 1.0
-16.7 ± 0.9
-18.3 ± 0.7
-18.1 ± 0.4
J/mol, K
-13.2 ± 3.0
-43.4 ± 0.4
-109.5 ± 3.3
-4.5 ± 2.9
-13.4 ± 2.4
-11.5 ± 1.4
kJ/mol
0
-10.9
-23.8
+4.1
+2.6
+2.8
296
kJ/mol
0
-8.9
-28.5
+2.6
-0.1
+0.5
K
1570
730
410
3740
1360
1570
Errors were calculated from the standard errors of the linear regression of Equation 7 and E was calculated from
the linear relation.
Role of the protein
There are a number of ways in which the protein can be affected to influence enzyme
enantioselectivity. Reports of changes in E caused by different immobilization techniques
[100], enzyme preparation [81] and mutagenesis [101] are all found in literature. Some may
have other causes than actual changes on the protein structure, such as diffusion limitations
within the catalyst particles of substrates to the active site [102]. Thermodynamic analysis of
observed changes in E with changes on the protein are, however, not common. More
information and conclusions may be drawn from the thermodynamic components than from
simply studying changes in the enantiomeric ratio.
Thermodynamic analysis
Overbeeke et al. performed thermodynamic analyses of observed changes in E with a
series of CALB-preparations chemically modified by covalent attachment of MPEG
(polyethylene glycol monomethyl ether), OPEG (polyethylene glycol monooctyl ether) and
1-octanol [85]. These modifications were shown to affect E and its differential activation
parameters for the kinetic resolution of 3-methyl-2-butanol with vinyl butyrate in hexane.
There are discrepancies between the activation parameters for native CALB found in that
study and in paper II. A possible explanation to this is that Overbeeke et al. used a
lyophilized enzyme preparation whereas in paper II an immobilized commercial preparation
was used. In their study the native CALB preparation and two preparations with different
extent of MPEG modification display negative differential activation enthalpy but positive
28
prot
ein
ce
rfa
su
acy
l sit
e
S105
L
alcoh
ol site
M
T42
W104
nucleophilic
elbow
T103
S47
Figure 17. Positions of the residues in Candida antarctica lipase B which were
subjected to point mutations in paper III. The preferred enantiomer of a sec-alcohol
is displayed in its reactive binding mode [60].
differential activation entropy. This is the unusual case of one enantiomer being favored both
by enthalpy and entropy. OPEG and 1-octanol modified CALB on the other hand display
counteracting differential activation parameters. The study proposes no clear correlation of
any structural or physical parameters of the modifier but an enthalpy-entropy compensation of
DR-SDH‡ and DR-SDS‡ is claimed.
Variations on the protein, in the form of one or two point mutations, were studied in
paper III for their effect on enantioselectivity and its thermodynamic components, Table 5.
E was drastically changed for better (T103G, E=2100) or worse (W104H, E=150) in the
resolution of 3-methyl-2-butanol compared to that catalyzed by wild type CALB (E=1000).
The underlying changes in the differential activation parameters, that caused these changes in
E, displayed a compensatory nature, as DR-SDH‡ and DR-SDS‡ regularly do. However, the
compensation was not perfect, which allowed the possibility to modify E. Although all
changes in E caused by the point mutations were caused by simultaneous changes in both
DR-SDH‡ and DR-SDS‡ it is noteworthy that for both T103G and W104H, which showed
opposite effects on E, the absolute values of DR-SDH‡ and DR-SDS‡ increased significantly.
Increased absolute values of the differential activation parameters as observed for these two
variants must be caused by structural changes that either enthalpically favored or entropically
disfavored the R-enantiomer transition state, vice versa for the S-enantiomer or a combination.
29
The increase in the absolute value of DR-SDH‡, for the T103G variant, was not to the full extent
compensated by the increase in DR-SDS‡ resulting in the increase in E. On the other hand for
W104H the increase in DR-SDH‡ was more than compensated by the increase in DR-SDS‡
resulting in a dramatic decrease in enantioselectivity. Models that exclude the contribution of
entropy would thus predict an erroneously high E-value for the W104H variant.
T103G constitutes part of what is usually termed the nucleophilic elbow, Figure 17,
and the mutation reintroduces the consensus sequence, GXSXG, common to many other
lipases. Possibly, the mutation brought about structural changes in the active site that
decreased the size of the stereospecificity pocket. Residue W104 is positioned in the bottom
of the active site and makes up part of the boundaries of the alcohol binding site and the
stereospecificity pocket. The mutation of this residue to a histidine liberated a large volume in
the active site. Whether this caused structural changes or left a free volume in the active site is
not known until the structure is determined. The dramatic drop in E may be interpreted as if
the mutation increased the volume of the stereospecificity pocket and hence reduced the steric
differences between the enantiomers. This is, however, a premature conclusion as observed in
the changes in the differential activation parameters. The actual difference in transition state
enthalpy has in fact increased and not decreased, as one might have guessed from the
observation of E alone. The structural change has in some way caused the R-enantiomer to be
even more favored by enthalpy and disfavored by entropy or vice versa for the S-enantiomer.
Rotticci et al. designed three variants T42V, S47A and T42V/S47A, for the
improvement of CALB enantioselectivity to halohydrins [97]. In these variants hydroxyl
groups within the active site stereospecificity pocket have been removed in exchange for a
methyl and a hydrogen. Repulsive interactions between these hydroxyl groups and the
halogens were proposed to be removed in the variants and hence halohydrins with halogens
on their medium-sized substituent were expected to be more successfully resolved [96, 97]. In
Table 6 one can note that the S47A and T42V/S47A variants did in fact show increased
enantioselectivity towards 1-bromo-2-butanol and 1-chloro-2-butanol whereas T42V showed
no effect (Paper III and [97]). The intermediate E values expressed by the double mutation
variant T42V/S47A indicated that the effects were not simply additive. The resolution of
3-methyl-2-butanol by these variants showed decreased enantioselectivity. On closer
inspection of the thermodynamic components the complex nature of enantioselectivity and the
possible erroneous conclusions one may draw from changes in E alone become evident. The
mutation, T42V, that had the least effect on E for 3-methyl-2-butanol, 1-bromo-2-butanol and
30
Table 6. Enantioselectivity, E, of Candida antarctica lipase B and variants (paper III).
Enzyme
E
1-bromo-21-chloro-23-methyl-2a
a
octanol
octanol
butanolb
296 K
296 K
296 K
Wild type
970
6.5 ± 0.4
14 ± 2
T42V
520
6.7 ± 0.3
15 ± 2
S47A
340
12.4 ± 0.4
28 ± 3
T42VS47A
390
9.4 ± 0.8
20 ± 2
Errors are one standard deviation. E for 3-methyl-2-butanol was calculated from the linear relation of Equation 7.
a) Transesterified with vinyl butanoate.
b) Transesterified with vinyl octanoate.
1-chloro-2-butanol displayed the largest changes in the differential activation parameters
compared to wild type, Table 5. Again these changes are hidden in E due to the compensatory
nature of enthalpy and entropy.
The study on the effect of variation on the protein (paper III) stress the importance of
entropy in enantioselectivity and the importance of not drawing premature conclusions from E
alone or from enzyme models that exclude the consideration of entropy. It is encouraging to
see that although enthalpy and entropy have a compensatory nature, this compensation is not
perfect, allowing for protein modifications to improve enantioselectivity.
The popular random (combinatorial) approaches of protein engineering for improved
properties of proteins [15] in combination with rational approaches and analysis of their
results will increase the understanding of protein structure-function relationships immensely.
Possibly, the prospect of general rational design of new and improved biocatalysts does not lie
far ahead into the future.
Role of the solvent
It is well recognized that enzyme performance is strongly affected by the choice of
solvent. Even reversal of substrate specificity [103, 104] and enantiopreference [105, 106] has
been observed by a change of solvent. The idea of solvent engineering is very appealing as the
solvent is an easily controlled reaction parameter. Consequently, much research have been
performed aimed to elucidate and understand the mechanisms of solvent effects on enzyme
reactions. Several rationales for solvent effects on enzyme enantioselectivity have been
proposed, but none has shown general applicability.
Many enzymes are affected by the water activity of the reaction system [107-109]. It
seems that enzymes require a degree of hydration to remain active and that their flexibility
31
and ability to accommodate the substrates are affected by water activity. However, CALB
seems to be little affected by variations in water activity in the resolution of
sec-alcohols ([58, 110], unpublished data).
The hydrophobicity of the solvent is another well-studied parameter for solvent effects
on enzyme enantioselectivity. Several studies report decreasing E with an increase in solvent
logP, where P is the partition coefficient between octanol and water [111-115]. This is
however not a general observation [26, 116-119]. Dipole moment and dielectricity constant
are other solvent parameters that have been tested for and claimed to have correlations
to E [116].
One model, based on molecular modeling, for rationalization of solvent effects on E
has been developed in the group of Klibanov. It is based on the assumption that solvent effects
are caused by differential desolvation free energy of the enzyme-substrate transition states.
The desolvated parts of the substrates are estimated through molecular modeling and their
thermodynamic activity coefficients are estimated with the UNIFAC algorithm [120].
Enantioselectivity and the ratio of the activity coefficients of the substrates are proposed to be
related according to logE = log(gR/gS) + constant. This model has been successful for some
reaction systems [121, 122] but neither this model has shown general applicability [123, 124].
None of the mentioned models view the solvent on a molecular level, instead the
solvent properties are treated as a continuum. A full understanding of solvent effects lies yet
in the future and the effects are most certainly composed of several different interacting
mechanisms [125]. Hopefully, thermodynamic analysis, such as in paper V, can bring further
light on this matter.
Thermodynamic analysis
In 1996 Noritomi et al. showed that the solvent influences the temperature dependence
of enantioselective enzymatic reactions [81]. In that study, E for the transesterification of secphenethyl alcohol catalyzed by Rhizomucor miehei lipase and subtilisin Carlsberg was studied
at two and three temperatures respectively. Depending on the solvent, increasing E with
temperature (Texp>Tr), decreasing E (Texp<Tr or Tr<0), and practically unaffected E (DDH‡»0)
were observed. If their data are plotted according to relative Arrhenius plots and the
differential activation parameters are determined, one notes that the changes in E were caused
by changes in both DDH‡ as well as DDS‡. In the same study nine different enzyme
preparations of subtilisin were tested and were shown to have very different E and
32
Table 7. Enantiomeric ratio, E, and its thermodynamic components for Candida antarctica
lipase B catalyzed transesterifications of 3-methyl-2-butanol with vinyl octanoate in nonaqueous solvents (paper V).
Solvent
Volume
E
∆R-S∆G‡
∆R-S∆H‡
T∆R-S∆S‡
∆R-S∆S‡
Tr
298 K
298 K
298 K
Å3
kJ/mol
kJ/mol
kJ/mol
J/mol,K
K
154
890
-16.8
950
Decaline
-25 ± 2.4 -7.7 ± 2.3 -26 ± 7.5
109
810
-16.6
940
Hexane
-24 ± 1.1 -7.7 ± 1.1 -26 ± 3.4
82
820
-16.6
Cyclopentane
-23 ± 3.1 -6.5 ± 3.1 -22 ±10.2 1060
80
580
-15.8
1,4-dioxane
-18 ± 0.3 -2.2 ± 0.3
-7.3± 1.1 2460
74
710
-16.3
Tetrahydrofuran
-20 ± 0.6 -3.2 ± 0.7 -11 ± 2.1 1790
60
650
-16.0
Acetone
-18 ± 1.2 -1.7 ± 1.2
-5.7± 4.0 3090
50
580
-15.8
Dichloromethane
-21 ± 1.6 -4.8 ± 1.7 -16 ± 5.5 1270
39
600
-15.9
Carbon disulfide
-20 ± 1.7 -4.4 ± 1.7 -15 ± 5.6 1370
a
a
26
330
-14.4
SCCO2
-19 ± 1.6 -4.7 ± 1.4a -16 ± 4.7 1210
Errors were calculated from the standard errors of the linear regression of Equation 7 and E was calculated from
the linear relation.
a
) Value calculated with extrapolation out of the experimental temperature range.
temperature dependence of E. One should bear in mind that with only two or three
temperatures linearity of Arrhenius plots and thus the relevancy of the model (Equation 6 or
7) cannot be verified. Furthermore, diffusional limitations may yield apparent E-values lower
than the intrinsic E of the enzyme [102].
Simpson et al. [126] observe that E and the temperature dependence of E in the
oxidation of 2-butanol catalyzed by a secondary alcohol dehydrogenase varies with the
amount of added cosolvent, acetonitrile. They make no analysis of the differential activation
parameters, probably because only three temperatures are studied. Rough estimation of DDH‡
and DDS‡ on their data, however, indicated that the changes in E upon cosolvent addition were
caused by changes in both differential activation enthalpy and entropy.
Complete thermodynamic analyses of changes in E with solvent composition for four
enzyme reaction systems were presented by Overbeeke et al. [127]. Changes in cosolvent
addition to both aqueous and non-aqueous media systems were studied. All systems were
below the racemic temperature and hence displayed increasing E with decreasing temperature.
The variation of E upon changes in solvent composition varied. Maxima, minima and
continually increasing values were observed. The underlying changes in the differential
activation parameters to these variations in E lie on both DR-SDH‡ and DR-SDS‡. The changes on
both components were large, but as they are counteracting when combined to the difference in
Gibbs free energy the effect on E was reduced.
33
1600
1400
SCCO
SCCO2
2
Carbon disulfide
Dichloromethane
Acetone
Tetrahydrofuran
Dioxane
Cyclopentane
Hexane
Decaline
Hexane 21.4 MPa
1200
1000
E 800
600
400
200
0
0
20
40
60
80
100
Temperature (°C)
Figure 18. The enantiomeric ratio, E, as a function of temperature for the kinetic
resolution of 3-methyl-2-butanol catalyzed by Candida antarctica lipase B in a range
of non-aqueous solvents (paper V). Vinyl octanoate was used as the acyl donor.
1000
900
cis -decaline
800
Cyclopentane
700
Hexane
Tetrahydrofuran
Acetone
600
Carbon
disulfide Dichloromethane
E
500
(298 K)
1,4-dioxane
400
300
SCCO2
200
100
0
0
50
100
150
200
V (Å3)
Figure 19. The enantiomeric ratio, E, as a function of the size of the solvent
molecule for the kinetic resolution of 3-methyl-2-butanol catalyzed by Candida
antarctica lipase B (paper V). Vinyl octanoate was used as the acyl donor.
34
0
50
100
150
200
0
-5
-10
(kJ/mol) -15
-20
-25
-30
Solvent volume (Å3)
Figure 20. Differential activation free energy (◊), ∆R-S∆G‡, and its enthalpic (●),
∆R-S∆H‡, and entropic (■), T∆R-S∆S‡, components at 298 K for the enantioselective
transesterification of 3-methyl-2-butanol with vinyl octanoate catalyzed by Candida
antarctica lipase B in solvents of different molecular volumes (paper V). Error bars
are calculated from the standard errors of the linear regression of Equation 7.
Solvent effects on E, caused by changes of both differential activation enthalpy and
entropy were also observed in paper V. CALB-catalyzed kinetic resolution through
transesterification was studied in eight non-aqueous liquid media and supercritical carbon
dioxide (SCCO2), Table 7. The enantiomeric ratio varied significantly with the choice of
solvent at all temperatures studied, Figure 18, and a correlation of E with the molecular size of
the solvent molecules was found, Figure 19. In the model of Orrenius et al. [60] the
enantiomers have different orientations in transition state, which may result in very different
interactions between the solvent and the activated enzyme-substrate complex. Differential
solvation may thus contribute to the differential activation parameters. For instance, a
difference in the number of solvent molecules involved in the solvation of the transition state
could contribute to DDS‡. If so, the size of the solvent molecule would influence the value of
DDS‡. A solvent with large molecular volume will lose translational entropy of fewer solvent
molecules than one with smaller molecular volume when restricted in the active site. The
differential activation parameters determined in paper V indicated a stepwise change with the
solvent volume, Figure 20. The small volume solvents displayed intermediate absolute values
of the differential activation parameters, the medium-sized solvents small values and the
35
large-sized solvents had large values. An upper limit would be reached when the solvent size
is too large and there is no difference between the enantiomers in the number of solvent
molecules involved in solvation of the transition state. For hexane and cis-decaline, E and its
temperature dependence was practically identical which suggests that this upper limit had
been reached. The remaining entropic contribution may consist of differential conformational
freedom of the substrate or the protein. However, as there is also a correlation between solvent
size and hydrophobicity of the solvent we cannot rule out that the observed effects are related
to hydrophobicity until a large polar or small hydrophobic solvent have been tested. Recent
finding that ionic liquids function as solvents for enzymatic reactions are of interest in this
respect [128].
Enthalpy-entropy compensation
Enthalpy-entropy compensation phenomena have been observed many times
([129, 130], paper I). Both in the study of Overbeeke et al. [127] and in paper V a strikingly
good linear correlation was found between DR-SDH‡ and DR-SDS‡ of varying solvent
composition, Figure 21. In paper I it was shown that a general enthalpy-entropy
compensation relation for enantioselective systems is an artifact of the range of quantifiable
E-values. Several authors have pointed out that this compensation very often is an artifact
arising from statistical errors and not a real chemical compensation [131-133]. McBane
illustrates this elegantly by showing that extremely good linear correlations may be achieved
from a completely random data set [133]. The only possibility to improve the statistics is to
increase the temperature range of the data sampling. This is however normally impossible for
enzymatic systems as they are limited by enzyme inactivation at high temperatures and low
activity at low temperatures.
Krug et al. sets up procedures to identify any real chemical compensation from
artifactual statistic ones [134-136]. These procedures have, in part, been used on both the data
in paper V as well as in the study of Overbeeke et al. Although the procedures could not with
certainty verify that the compensation is of non-statistical origin there were at least strong
indications of a chemical compensation. In the study of Overbeeke et al. differential activation
enthalpy and entropy form hairpin shaped relations that they propose indicate the existence of
two thermodynamic states of the protein and that solvent changes affect the equilibrium
constant between these two states. One may note from the study in paper V that the reaction
catalyzed in SCCO2 did not follow the enthalpy-entropy compensation observed for the eight
36
-25
-23
-21
-19
-17
-5
60Å3
80Å3
‡
DDS (J/mol, K)
74Å3
39Å3
50Å3
-15
-20
82Å3
154Å3
109Å
26Å3
-10
-25
3
-30
‡
DDH (kJ/mol)
Figure 21. Differential activation entropy as a function of differential activation
enthalpy for the kinetic resolution of 3-methyl-2-butanol catalyzed by Candida
antarctica lipase B in a range of solvents of different molecular size (paper V).
Vinyl octanoate was used as the acyl donor. See Table 7 for identification of solvent
with its volume. R2 from a linear regression is 0.9904 if SCCO2 (26 Å3) is excluded
and 0.9367 when included.
liquid solvents, Figure 21. This implies that the reaction in the supercritical media was in
some way different in mechanism compared to the reaction in the condensed liquid media. As
the reaction catalyzed in pressurized hexane showed no variation in enantiomeric ratio
compared to the reaction in hexane at atmospheric pressure, the effect was not caused by the
increased pressure. The nature of this different mechanism is unclear but changes on the
protein structure caused by SCCO2 are possible. Carbon dioxide is known to covalently
modify lysine residues and it may associate with the hydration shell of the enzyme, lowering
the effective pH in the enzyme microenvironment.
37
Concluding remarks
Our understanding of enzyme specificity has advanced greatly from the static view of
Fischer’s “lock and key” mechanism. It is evident that both enzyme and substrate are dynamic
objects that accommodate each other. However, there is still much to learn about the structure
function relationships of enzymes before sufficient knowledge on a molecular level is
acquired to allow rational design of enzymes. For enantioselective enzymes a better
understanding of the roles of differential activation enthalpy and entropy, in particular, may
enable such rational design. Entropy is for most of us an abstract concept although strictly
defined, or in fact postulated, as a state function in the second law of thermodynamics. The
activation entropy of enzyme reactions is certainly composed of several contributions.
Solvation, loss of translational, rotational and vibrational degrees of freedom for the substrate
and enzyme residues all contribute to the activation entropy. Separation of the various
components is practically impossible since changing one parameter invariably changes the
conditions for the other parameters simultaneously. The compensative character of enthalpy
and entropy further complicates the elucidation of their individual roles. The advances of
handling activation entropy within molecular modeling [137, 138] are however promising for
the development of complete models of enzymatic systems that take both enthalpy and
entropy into account. This may bring forth the possibility not only of rational design for
improvement of enzymes for already existing reactions but possibly be of value for the design
of catalysts for new reactions.
38
Acknowledgements
I would like to express my sincere gratitude to all involved in the making of this thesis.
I am especially grateful to my supervisor Professor Karl Hult for his support, encouragement
and contagious enthusiasm to science. Also for all the non-scientific excursions he has guided
on skates, skis, rafts and foot that make fine memories.
All the collaborators in the European Union project (grant number BIO4-CT95-0231) from
Milan and Delft are thanked for rewarding and fruitful discussions and collaborations.
I am indebted to the members of the biocatalysis group that provided invaluable comments to
the manuscript of this thesis.
All past and present members of the biocatalysis group are thanked for providing a friendly
and stimulating working atmosphere. Professor Torbjörn Norin for his encouragement and
advice. Docent Per Berglund who has generously shared his expertise and vast literature
library. My constant office and lab companion Fredrik Viklund for his patient support in many
matters. Linda Fransson for fun collaboration and her untiring and generous help with pictures
and comments to the papers and the thesis. My coauthors Joke Rotticci-Mulder and Didier
Rotticci for all help and exciting collaborations.
Special thanks to Dr. Jerry King, my supervisor and mentor in the field of supercritical fluids.
I am most grateful to the members and collaborators of his supercritical fluids group at
NCAUR in Peoria, Illinois. Jeff Teel is varmly thanked for his invaluable technical expertise
and generous help.
Thanks to Janne Lehtiö for your help in our parallel race against the clock, and for putting me
in the nice position of always having two more weeks.
I would also like to thank all colleagues and friends at the department of Biotechnology for an
enjoyable time during the years.
Finally I would also like to thank all friends and a large family, especially:
Ann, Dan
My goddaughter Klara for giving life perspective
Peter
My brother and sister and best friends, Anna and John
My parents, Anita and Henry, for their unlimited support and encouragement at all times.
39
References
1.
P. Ball, K. Ziemelis, L. Allen, Biocatalysis: Synthesis methods that exploit enzymatic
activities, Nature 409 (2001) 225-225.
2.
J. Tramper, Applied biocatalysis: from product request to idea to product, in Applied
biocatalysis, J.M.S. Cabral, et al., Editors. 1993, Harwood Academic. p. 1-45.
3.
K. B. Buchholz, P. B. Poulsen, Introduction, in Applied biocatalysis, A.J.J. Straathof
and P. Adlercreutz, Editors. 2000, Harwood Academic: Amsterdam. p. 1-15.
4.
L. Stryer, Biochemistry. 4 ed. 1995, New York: Freeman.
5.
S. L. Neidleman, Selected historical perspectives in biocatalysis, 1752-1960, in ACS
Symp. Ser. Applied biocatalysis in speciality chemicals and pharmaceuticals, B.C. Saha
and D.C. Demirjian, Editors. 2001. p. 14-40.
6.
A. C. Hill, Reversible zymohydrolysis J. Chem. Soc. 73 (1898) 634-658.
7.
J. H. Kastle, A. S. Loevenhart, Concerning lipase, the fat-splitting enzyme, and the
reversibility of its action, Am. Chem. J. 24 (1900) 491-525.
8.
P. Halling, L. Kvittingen, Why did biocatalysis in organic media not take of in the
1930s?, Trends Biotechnol. 17 (1999) 343-344.
9.
A. Fersht, Enzyme structure and mechanism. 2 ed. 1985, New York: Freeman.
10. A. Schmid, J. S. Dordick, B. Hauer, A. Kiener, M. Wubbolts, B. Witholt, Industrial
biocatalysis today and tomorrow, Nature 409 (2001) 258-268.
11.
C. Walsh, Enabling the chemistry of life, Nature 409 (2001) 226-231.
12. K. M. Koeller, C. H. Wong, Enzymes for chemical synthesis, Nature 409 (2001) 232240.
13. A. M. Klibanov, Improving enzymes by using them in organic solvents, Nature 409
(2001) 241-246.
14. C. Khosla, P. B. Harbury, Modular enzymes, Nature 409 (2001) 247-252.
15.
F. H. Arnold, Combinatorial and computational challenges for biocatalyst design,
Nature 409 (2001) 253-257.
16. A. M. Thayer, Biocatalysis, Chem. Eng. News 79 (2001) 27-34.
17.
M. McCoy, Making drugs with little bugs, Chem. Eng. News 79 (2001) 37-43.
40
18. J. F. Tremblay, Japan's unique perspective, Chemical & Engineering News 79 (2001)
45-49.
19. E. Antonini, G. Carrea, P. Cremonesi, Enzyme catalysed reactions in water-organic
solvent two-phase systems, Enzyme Microb. Technol. 3 (1981) 291-296.
20. K. Martinek, A. N. Semenov, I. V. Berezin, Enzymatic synthesis in biphasic aqueousorganic systems I. Chemical equilibrium shift, Biochim. Biophys. Acta 658 (1981) 7689.
21. K. Martinek, A. V. Levashov, Y. L. Khmelnitsky, N. L. Klyachko, I. V. Berezin,
Colloidal solution of water in organic solvents: a microheterogeneous medium for
enzymatic reactions, Science 218 (1982) 889-891.
22. A. Zaks, A. M. Klibanov, Enzymatic catalysis in organic media at 100°C, Science 224
(1984) 1249-1251.
23. A. Zaks, A. M. Klibanov, Enzyme-catalyzed processes in organic solvents, Proc. Natl.
Acad. Sci. U. S. A. 82 (1985) 3192-3196.
24. J. Partridge, B. D. Moore, P. J. Halling, a-Chymotrypsin stability in aqueous-acetonitrile
mixtures: is the native enzyme thermodynamically or kinetically stable under low water
conditions?, J. Mol. Catal. B: Enzym. 6 (1999) 11-20.
25. K. Griebenow, A. M. Klibanov, On protein denaturation in aqueous-organic mixtures
but not in pure organic solvents, J. Am. Chem. Soc. 118 (1996) 11695-11700.
26. C. R. Wescott, A. M. Klibanov, The solvent dependence of enzyme specificity, Biochim.
Biophys. Acta 1206 (1994) 1-9.
27. A. Zaks, A. M. Klibanov, Enzymatic catalysis in nonaqueous solvents, J. Biol. Chem.
263 (1988) 3194-3201.
28.
A. Persidis, Extremophiles, Nat. Biotechnol. 16 (1998) 593-594.
29. M. W. W. Adams, R. M. Kelly, Enzymes from microorganisms in extreme
environments, Chem. Eng. News 73 (1995) 32-42.
30. K. Faber, Biotransformations in organic chemistry. 4 ed. 2000, Berlin: Springer.
31. T. W. Randolph, Blanch, H. W., Prausnitz, J. M., Wilke, C. R., Enzymatic catalysis in a
supercritical fluid, Biotechnol. Lett. 7 (1985) 325-328.
32. D. A. Hammond, Karel M., Klibanov, A. M., Enzymatic reactions in supercritical gases,
Appl. Biochem. Biotechnol. 11 (1985) 393-400.
33. A. J. Mesiano, E. J. Beckman, A. J. Russel, Supercritical biocatalysis, Chem. Rev. 99
(1999) 623-633.
41
34.
E. Cernia, C. Palocci, Lipases in supercritical fluids, Methods in Enzymology 286 (1997)
495-508.
35. S. V. Kamat, E. J. Beckman, A. J. Russel, Enzyme activity in supercritical fluids, Crit.
Rev. Biotechnol. 15 (1995) 41-71.
36. R. D. Schmid, R. Verger, Lipases: interfacial enzymes with attractive applications,
Angew. Chem. Int. Ed. 37 (1998) 1608-1633.
37.
P. Berglund, K. Hult, Biocatalytic synthesis of enantipoure compounds using lipases, in
Stereoselective Biocatalysis, R.N. Patel, Editor. 2000, Marcel Dekker, Inc.: New York,
Basel. p. 633-657.
38. G. W. Muller, Thalidomide: From tragedy to new drug discovery, CHEMTECH 27
(1997) 21-25.
39.
S. C. Stinson, Chiral drugs, Chem. Eng. News 78 (2000) 55-78.
40.
S. C. Stinson, Counting on chiral drugs, Chem. Eng. News 76 (1998) 83-104.
41. G. P. Moss, Basic terminology of stereochemistry, Pure Appl. Chem. 68 (1996)
2193-2222.
42.
S. Schurig, H.-P. Nowotny, Separation of enantiomers on diluted permethylated bcyclodextrin by high-resolution gas chromatography, J. Chromatogr. 441 (1988) 155163.
43. M. McCoy, Biocatalysis grows for drug synthesis, Chem. Eng. News 77 (1999) 10-14.
44. C. R. Johnson, G. W. Wells, Organic synthesis using biocatalytically generated
intermediates, Curr. Opin. Chem. Biol. 2 (1998) 70-76.
45. A. Liese, M. V. Filho, Production of fine chemicals using biocatalysis, Curr. Opin.
Biotechnol. 10 (1999) 595-603.
46.
C.-S. Chen, Y. Fujimoto, G. Girdaukas, C. J. Sih, Quantitative analysis of biochemical
kinetic resolutions of enantiomers, J. Am. Chem. Soc. 104 (1982) 7294-7299.
47. H. W. Anthonsen, B. H. Hoff, T. Anthonsen, Calculation of enantiomer ratio and
equilibrium constants in biocatalytic ping-pong bi-bi resolutions, Tetrahedron:
Asymmetry 7 (1996) 2633-2638.
48. H. W. Anthonsen, B. H. Hoff, T. Anthonsen, A simple method for calculating
enantiomer ratio and equilibrium constants in biocatalytic resolutions, Tetrahedron:
Asymmetry 6 (1995) 3015-3022.
49. J. L. L. Rakels, A. J. J. Straathof, J. J. Heijnen, A simple method to determine the
enantiomeric ratio in enantioselective biocatalysis, Enzyme Microb. Technol. 15 (1993)
1051-1056.
42
50. A. J. J. Straathof, J. A. Jongejan, The enantiomeric ratio: origin, determination and
prediction, Enzyme Microb. Technol. 21 (1997) 559-571.
51. C.-S. Chen, S.-H. Wu, G. Girdaukas, C. J. Sih, Quantitative analyses of biochemical
kinetic resolution of enantiomers. 2. Enzyme-catalyzed esterifications in water-organic
solvent biphasic systems, J. Am. Chem. Soc. 109 (1986) 2812-2817.
52.
P. Berglund, Controlling lipase enantioselectivity for organic synthesis, Biomol. Eng. 18
(2001) 13-22.
53. E. M. Anderson, K. M. Larsson, O. Kirk, One biocatalyst - many applications: the use of
Candida antarctica B-lipase in organic synthesis, Biocatal. Biotransform. 16 (1998)
181-204.
54. J. Uppenberg, M. T. Hansen, S. Patkar, T. A. Jones, The sequence, crystal structure
determination and refinement of two crystal forms of lipase B from Candida antarctica,
Structure 2 (1994) 293-308.
55. D. L. Ollis, E. Cheah, M. Cygler, B. Dijkstra, F. Frolow, S. M. Franken, M. Harel, S. J.
Remington, I. Silman, J. Schrag, J. L. Sussman, K. H. G. Verschueren, A. Goldman, The
a/b hydrolase fold, Protein Eng. 5 (1992) 197-211.
56. M. Martinelle, M. Holmquist, K. Hult, On the interfacial activation of Candida
antarctica lipase A and B as compared with Humicola lanuginosa lipase, Biochim.
Biophys. Acta 1258 (1995) 272-276.
57. R. J. Kazlauskas, A. N. E. Weissfloch, A. T. Rappaport, L. A. Cuccia, A rule to predict
which enantiomer of a secondary alcohol reacts faster in reactions catalyzed by
cholesterol esterase, lipase from Pseudomonas cepacia, and lipase from Candida
rugosa, J. Org. Chem. 56 (1991) 2656-2665.
58.
D. Rotticci, J. Ottosson, T. Norin, K. Hult, Candida antarctica lipase B: a tool for the
preparation of optically active alcohols, in Enzymes in Nonaqueous Solvents: Methods
and Protocols, V.E. N., H.P. J., and H.H. L., Editors. 2001, Humana Press: Totowa. p.
261-276.
59. J. Uppenberg, N. Öhrner, M. Norin, K. Hult, G. J. Kleywegt, S. Patkar, V. Waagen, T.
Anthonsen, T. A. Jones, Crystallographic and molecular-modeling studies of Lipase B
from Candida antarctica revael a stereospecificity pocket for secondary alcohols,
Biochemistry 34 (1995) 16838-16851.
60. C. Orrenius, F. Hæffner, D. Rotticci, N. Öhrner, T. Norin, K. Hult, Chiral recognition of
alcohol enantiomers in acyl transfer reactions catalysed by Candida antarctica lipase B,
Biocatal. Biotransform. 16 (1998) 1-15.
61. R. S. Philips, Temperature effects on stereochemistry of enzymatic reactions, Enzyme
Microb. Technol. 14 (1992) 417-419.
62. H. M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T. N. Bhat, H. Weissig, I. N.
Shindyalov, The protein data bank, Nucleic Acids Res. 28 (2000) 235-242.
43
63. A. R. Leach, Molecular modelling. 2 ed. 2001, Harlow: Prentice Hall.
64. R. J. Kazlauskas, Molecular modeling and biocatalysis: explanations, predictions,
limitations, and opportunities, Curr. Opin. Chem. Biol 4 (2000) 81-88.
65.
E. N. Vulfson, P. J. Halling, H. L. Holland, eds. Enzymes in nonaqueous solvents:
methods and protocols. Methods in Biotechnology, ed. J.M. Walker. 2001, Humana
Press: Totowa.
66. U. Bornscheuer, R. J. Kazlauskas, Hydrolases in organic synthesis. 1999, Weinheim:
Wiley-VCH.
67. D. F. DeTar, Computation of enzyme-substrate specificity, Biochemistry 20 (1981)
1730-1743.
68. E. Holmberg, K. Hult, Temperature as an enantioselective parameter in enzymatic
resolutions of racemic mixtures, Biotechnol. Lett. 13 (1991) 323-326.
69. M. Norin, K. Hult, A. Mattson, T. Norin, Molecular modelling of chymotrypsinsubstrate interactions: calculation of enantioselectivity, Biocatalysis 7 (1993) 131-147.
70.
V. T. Pham, R. S. Phillips, L. G. Ljungdahl, Temperature-dependent enantiospecificity
of secondary alcohol dehydrogenase from Thermoanaerobacter ethanolicus, J. Am.
Chem. Soc. 111 (1989) 1935-1936.
71.
R. S. Philips, Temperature modulation of the stereochemistry of enzymatic catalysis:
prospects for exploitation, Trends Biotechnol. 14 (1996) 13-16.
72. T. Miyazawa, K. Imagawa, R. Yanagihara, T. Yamada, Marked dependence on
temperature of enantioselectivity in the Aspergillus oryzae protease-catalyzed hydrolysis
of amino acid esters, Biotechnol. Tech. 11 (1997) 931-933.
73. B. Galunsky, S. Ignatova, V. Kasche, Temperature effects on S1- and S'1enantioselectivity of a-chymotrypsin, Biochim. Biophys. Acta 1343 (1997) 130-138.
74. H. Yang, Å. Jönsson, E. Wehtje, P. Adlercreutz, B. Mattiasson, The enantiomeric purity
of alcohols formed by enzymatic reduction of ketones can be improved by optimisation
of the temperature and by using a high co-substrate concentration, Biochim. Biophys.
Acta 1336 (1997) 51-58.
75.
T. Miyazawa, T. Yukawa, S. Ueji, R. Yanagihara, T. Yamada, Resolution of 2-phenoxy1-propanols by Pseudomonas sp. lipase- catalyzed highly enantioselective
transesterification: influence of reaction conditions on the enantioselectivity toward
primary alcohols, Biotechnol. Lett. 20 (1998) 235-238.
76. Å. Jönsson, E. Wehtje, P. Adlercreutz, B. Mattiasson, Thermodynamic and kinetic
aspects on water vs. organic solvent as reaction media in the enzyme-catalysed reduction
of ketones, Biochim. Biophys. Acta 1430 (1999) 313-322.
44
77. P. Lopez-Serrano, M. A. Wegman, F. van Rantwijk, R. A. Sheldon, Enantioselective
enzyme catalysed ammoniolysis of amino acid derivatives. Effect of temperature,
Tetrahedron: Asymmetry 12 (2001) 235-240.
78. K. Watanabe, T. Koshiba, Y. Yasufuku, T. Miyazawa, S. Ueji, Effects of substituent and
temperature on enantioselectivity for lipase-catalyzed esterification of 2-(4-substituted
phenoxy) propionic acids in organic solvents, Bioorg. Chem. 29 (2001) 65-76.
79. R. Bauer, H. J. Knackmuss, A. Stolz, Enantioselective hydration of 2-arylpropionitriles
by a nitrile hydratase from Agrobacterium tumefaciens strain d3, Appl. Microbiol.
Biotechnol. 49 (1998) 89-95.
80. Y. Yasufuku, S. Ueji, High temperature-induced high enantioselectivity of lipase for
esterifications of 2-phenoxypropionic acids in organic solvent, Bioorg. Chem. 25 (1997)
88-99.
81. H. Noritomi, Ö. Älmarsson, G. L. Barletta, A. M. Klibanov, The influence of the mode
of enzyme preparation on enzymatic enantioselectivity in organic solvents and its
temperature dependence, Biotechnol. Bioeng. 51 (1996) 95-99.
82. Y. Yasufuku, S.-i. Ueji, Effect of temperature on lipase-catalyzed esterification in
organic solvent, Biotechnol. Lett. 17 (1995) 1311-1316.
83. W. H. Lee, K.-J. Kim, M. G. Kim, S. B. Lee, Enzymatic resolution of racemic ibuprofen
esters: effects of organic cosolvents and temperature, J. Ferm. Bioeng. 80 (1995) 613615.
84. V. T. Pham, R. S. Phillips, Effects of substrate structure and temperature on the
stereospecificity of secondary alcohol dehydrogenase from Thermoanaerobacter
ethanolicus, J. Am. Chem. Soc. 112 (1990) 3629-3632.
85. P. L. A. Overbeeke, B. C. Koops, H. M. Verheij, A. J. Slotboom, M. R. Egmond, J. A.
Jongejan, J. J. Heijnen, Activity and enantioselectivity of modified lipases in organic
solvents, Biocatal. Biotransform. 18 (2000) 59-77.
86. J. B. Jones, I. J. Jakovac, A new cubic-space section model fpr predicting the specificity
of horse liver alcohol dehydrogenase-catalyzed oxidoreductions, Can. J. Chem. 60
(1981) 19-28.
87. E. J. Toone, M. J. Werth, J. B. Jones, Active-site model for interpreting and predicting
the specificity of pig liver esterase, J. Am. Chem. Soc. 112 (1990) 4946-4952.
88. K. Naemura, R. Fukuda, M. Konishi, K. Hirose, Y. Tobe, Lipase YS-catalyzed acylation
of alcohols - a predictive active-site model for lipase YS to identify which enantiomer of
a primary or a secondary alcohol reacts faster in this acylation, J. Chem. Soc., Perkin
Trans. 1 (1994) 1253-1256.
89. G. Ottolina, G. Carrea, S. Colonna, A. Ruckemann, A predictive active site model for
cyclohexanone monooxygenase catalyzed Baeyer-Villiger oxidations, TetrahedronAsymmetry 7 (1996) 1123-1136.
45
90. G. Wipff, A. Dearing, P. K. Weiner, J. M. Blaney, P. A. Kollman, Molecular mechanics
studies of enzyme-substrate interactions: the interaction of L- and D-Nacetyltryptophaneamide with a-Chymotrypsin, J. Am. Chem. Soc. 105 (1983).
91. B. Y. Hwang, H. Scheib, J. Pleiss, B. G. Kim, R. D. Schmid, Computer-aided molecular
modeling of the enantioselectivity of Pseudomonas cepacia lipase toward g- and dlactones, J. Mol. Catal. B: Enzym. 10 (2000) 223-231.
92.
T. Schulz, J. Pleiss, R. D. Schmid, Stereoselectivity of Pseudomonas cepacia lipase
toward secondary alcohols: A quantitative model, Protein Sci. 9 (2000) 1053-1062.
93.
S. Raza, L. Fransson, K. Hult, Enantioselectivity in Candida antarctica lipase B: A
molecular dynamics study, Protein Sci. 10 (2001) 329-338.
94. T. Ke, A. M. Klibanov, Markedly enhancing enzymatic enantioselectivity in organic
solvents by forming substrate salts, J. Am. Chem. Soc. 121 (1999) 3334-3340.
95.
C. S. Zheng, R. S. Phillips, Effect of coenzyme analogs on enantioselectivity of alcohol
dehydrogenase, J. Chem. Soc., Perkin Trans. 1 (1992) 1083-1084.
96. D. Rotticci, C. Orrenius, K. Hult, T. Norin, Enantiomerically enriched bifunctional secalcohols prepared by Candida antarctica Lipase B catalysis. Evidence of non-steric
interactions, Tetrahedron: Assymetry 8 (1997) 359-362.
97. D. Rotticci, J. C. Rotticci-Mulder, S. Denman, T. Norin, K. Hult, Improved
enantioselectivity of a lipase by rational protein engineering, ChemBioChem (2001) In
press.
98. F. Haeffner, T. Norin, K. Hult, Molecular modeling of the enantioselectivity in lipasecatalyzed transesterification reactions, Biophys. J. 74 (1998) 1251-1262.
99. P. J. Halling, Biocatalysis in low-water media: understanding effects of reaction
conditions, Curr. Opin. Chem. Biol. 4 (2000) 74-80.
100. G. Fernandez-Lorente, M. Terreni, C. Mateo, A. Bastida, R. Fernandez-Lafuente, P.
Dalmases, J. Huguet, J. M. Guisan, Modulation of lipase properties in macro-aqueous
systems by controlled enzyme immobilization: enantioselective hydrolysis of a chiral
ester by immobilized Pseudomonas lipase, Enzyme Microb. Technol. 28 (2001) 389-396.
101. M. T. Reetz, A. Zonta, K. Schimossek, K. Liebeton, K. E. Jaeger, Creation of
enantioselective biocatalysts for organic chemistry by in vitro evolution, Angew. Chem.,
Int. Ed. 36 (1997) 2830-2832.
102. D. Rotticci, T. Norin, K. Hult, Mass transport limitations reduce the effective
stereospecificity in enzyme-catalyzed kinetic resolution, Org. Lett. 2 (2000) 1373-1376.
103. A. Zaks, A. ö. M. Klibanov, Substrate specificity of enzymes in organic solvents vs.
water is reversed, J. Am. Chem. Soc. 108 (1986) 2767-2768.
46
104. C. R. Wescott, A. M. Klibanov, Solvent variation inverts substrate-specificity of an
enzyme, J. Am. Chem. Soc. 115 (1993) 1629-1631.
105. S. Tawaki, A. M. Klibanov, Inversion of enzyme enantioselectivity mediated by the
solvent, J. Am. Chem. Soc. 114 (1992) 1882-1884.
106. C. R. Wescott, A. M. Klibanov, Thermodynamic analysis of solvent effect on substrate
specificity of lyophilized enzymes suspended in organic media, Biotechnol. Bioeng. 56
(1997) 340-344.
107. P. J. Halling, Rates of enzymic reactions in predominantly organic low water systems,
Biocatalysis 1 (1987) 109-115.
108. P. J. Halling, Organic liquids and biocatalysts: theory and practice, Trends Biotechnol. 7
(1989) 50-52.
109. R. H. Valivety, P. J. Halling, A. D. Peilow, A. R. Macrae, Lipases from different sources
vary widely in dependence of catalytic activity on water activity, Biochim. Biophys. Acta
1122 (1992) 143-146.
110. E. Wehtje, D. Costes, P. Adlercreutz, Enantioselectivity of lipases: effects of water
activity, J. Mol. Catal. B: Enzym. 3 (1997) 221-230.
111. R. Morrone, G. Nicolosi, A. Patti, M. Piattelli, Resolution of racemic flurbiprofen by
lipase-mediated esterification in organic solvent, Tetrahedron: Asymmetry 6 (1995)
1773-1778.
112. A. Ducret, P. Pepin, M. Trani, R. Lortie, Lipase-catalyzed selective esterification of
ibuprofen, Ann. N. Y. Acad. Sci. 799 (1996) 747-751.
113. F. Roure, A. Ducret, M. Trani, R. Lortie, Enantioselective esterification of racemic
ibuprofen in solvent media under reduced pressure, J. Chem. Technol. Biotechnol. 69
(1997) 266-270.
114. K. Kawashiro, H. Sugahara, S. Sugiyama, H. Hayashi, Effect of organic solvents on
enantioselectivity of protease catalysis, Biotechnol. Bioeng. 53 (1997) 26-31.
115. A. Ducret, M. Trani, R. Lortie, Lipase-catalyzed enantioselective esterification of
ibuprofen in organic solvents under controlled water activity, Enzyme Microb. Technol.
22 (1998) 212-216.
116. P. A. Fitzpatrick, A. M. Klibanov, How can the solvent affect enzyme enantioselectivity,
J. Am. Chem. Soc. 113 (1991) 3166-3171.
117. C. Orrenius, T. Norin, K. Hult, G. Carrea, The Candida antarctica lipase B catalysed
kinetic resolution of seudenol in non-aqueous media of controlled water activity,
Tetrahedron: Asymmetry 6 (1995) 3023-3030.
118. R. Morrone, G. Nicolosi, A. Patti, Resolution of racemic 1-hydroxyalkylferrocenes by
lipase B from Candida antarctica, Gazz. Chim. Ital. 127 (1997) 5-9.
47
119. E. Forro, L. T. Kanerva, F. Fulop, Lipase-catalyzed resolution
dialkylaminomethylcyclohexanols, Tetrahedron: Asymmetry 9 (1998) 513-520.
of
2-
120. A. Fredenslund, R. L. Jones, J. M. Prausnitz, Group-contribution estimation of activity
coefficients in nonideal liquid mixtures, AlChE Journal 21 (1975) 1086-1099.
121. T. Ke, C. R. Wescott, A. M. Klibanov, Prediction of the solvent dependence of
enzymatic prochiral selectivity by means of structure-based thermodynamic
calculations, J. Am. Chem. Soc 118 (1996) 3366-3374.
122. C. R. Wescott, H. Noritomi, A. M. Klibanov, Rational control of enzymatic
enantioselectivity through solvation thermodynamics, J. Am. Chem. Soc. 118 (1996)
10365-10370.
123. G. Colombo, G. Ottolina, G. Carrea, A. Bernardi, C. Scolastico, Application of
structure-based thermodynamic calculations to the rationalization of the
enantioselectivity of subtilisin in organic solvents, Tetrahedron: Asymmetry 9 (1998)
1205-1214.
124. S. Luque, T. Ke, A. M. Klibanov, On the role of transition-state substrate desolvation in
enzymatic enantioselectivity in aqueous-organic mixtures, Biocatal. Biotransform. 16
(1998) 233-248.
125. G. Carrea, S. Riva, Properties and synthetic applications of enzymes in organic solvents,
Angew. Chem., Int. Ed. 39 (2000) 2226-2254.
126. H. D. Simpson, D. A. Cowan, Controlling the enantioselectivity of sec-alcohol
dehydrogenase from Thermoanaerobacterium sp. Ket4B1, Protein Pept. Lett. 4 (1997)
25-31.
127. P. L. A. Overbeeke, J. A. Jongejan, J. J. Heijnen, Solvent effect on lipase
enantioselectivity. Evidence for the presence of two thermodynamic states, Biotechnol.
Bioeng. 70 (2000) 278-290.
128. M. Erbeldinger, A. J. Mesiano, A. J. Russell, Enzymatic catalysis of formation of Zaspartame in ionic liquid - An alternative to enzymatic catalysis in organic solvents,
Biotechnol. Prog. 16 (2000) 1129-1131.
129. J. E. Leffler, The enthalpy-entropy relationship and its implications for organic
chemistry, J. Org. Chem. 20 (1955) 1202-1231.
130. Y. B. Tewari, M. M. Schantz, P. C. Pandey, M. V. Rekharsky, R. N. Goldberg,
Thermodynamics of the hydrolysis of N-acetyl-L-phenylalanine ethyl ester in water and
in organic solvents, J. Phys. Chem. 99 (1995) 1594-1601.
131. O. Exner, The enthalpy-entropy relationship, Prog. Phys. Org. Chem. 10 (1973) 411482.
132. O. Exner, Concerning the isokinetic relationship, Nature 201 (1964) 488-490.
48
133. G. C. McBane, Chemistry from telephone numbers: The false isokinetic relationship, J.
Chem. Educ. 75 (1998) 919-922.
134. R. R. Krug, W. G. Hunter, R. A. Grieger, Statistical interpretation of enthalpy-entropy
compensation, Nature 261 (1976) 566-567.
135. R. R. Krug, W. G. Hunter, R. A. Grieger, Enthalpy-entropy compensation. 1. Some
fundamental statistical problems associated with the analysis of van't Hoff and
Arrhenius data, J. Phys. Chem. 80 (1976) 2335-2341.
136. R. R. Krug, W. G. Hunter, R. A. Grieger, Enthalpy-entropy compensation. 2. Separation
of the chemical from the statistical effect, J. Phys. Chem. 80 (1976) 2341-2351.
137. J. Villa, M. Strajbl, T. M. Glennon, Y. Y. Sham, Z. T. Chu, A. Warshel, How important
are entropic contributions to enzyme catalysis?, Proc. Natl. Acad. Sci. U. S. A. 97
(2000) 11899-11904.
138. M. Strajbl, Y. Y. Sham, J. Villa, Z. T. Chu, A. Warshel, Calculations of activation
entropies of chemical reactions in solution, J. Phys. Chem. B 104 (2000) 4578-4584.
49