Molecular basis of chiral acid recognition by Candida rugosa lipase: X-ray structure of transition state analog and modeling of the hydrolysis of methyl 2-methoxy-2-phenylacetate

Article abstract of DOI:10.1002/adsc.201100459

Lipase from Candida rugosa shows high enantioselectivity toward α-substituted chiral acids such as 2-arylpropionic acids. To understand how Candida rugosa lipase (CRL) distinguishes between enantiomers of chiral acids, we determined the X-ray crystal stru

Full text of DOI:10.1002/adsc.201100459

FULL PAPERS
DOI: 10.1002/adsc.201100459
Molecular Basis of Chiral Acid Recognition by Candida rugosa
Lipase: X-Ray Structure of Transition State Analog and Modeling
of the Hydrolysis of Methyl 2-Methoxy-2-phenylacetate
Ian J. Colton,a,d DeLu (Tyler) Yin,b Pawel Grochulski,c,e,*
and Romas J. Kazlauskasa,b,*
a
McGill University, Department of Chemistry, 801 Sherbrooke Street West, Montrꢀal, Quꢀbec H3A 2K6, Canada
University of Minnesota, Department of Biochemistry, Molecular Biology & Biophysics and The BioTechnology Institute,
b
1479 Gortner Avenue, Saint Paul, MN 55108, USA
Fax : (+1)-612-625-5780; e-mail: rjk@umn.edu
National Research Council of Canada, Biotechnology Research Institute, 6100 Royalmount Avenue, Montrꢀal, Quꢀbec
c
H4P 2R2, Canada
d
Current address: Gowlings, 160 Elgin Street, Suite 2600, Ottawa, Ontario K1P 1C3, Canada
Current address: Canadian Light Source, University of Saskatchewan, 101 Perimeter Road, Saskatoon, SK S7N 0X4,
e
Canada
E-mail: pawel.grochulski@lightsource.ca
Received: June 5, 2011; Revised: August 26, 2011; Published online: September 5, 2011
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/adcs.201100459.
Abstract: Lipase from Candida rugosa shows high stereocenter. Computer modeling of the slow-react-
enantioselectivity toward a-substituted chiral acids ing enantiomer examined four possible conforma-
such as 2-arylpropionic acids. To understand how tions for the corresponding slow-reacting enantio-
Candida rugosa lipase (CRL) distinguishes between mer: three conformations where two substituents at
enantiomers of chiral acids, we determined the X-ray the stereocenter have been exchanged relative to the
crystal structure of a transition-state analog covalent- fast-reacting enantiomer and one conformation with
ly linked to CRL. CRL shows moderate enantiose- an umbrella-like inversion orientation. Each of these
lectivity (E=23) toward methyl 2-methoxy-2-phenyl- orientations disrupts the orientation of the catalytic
acetate, 1-methyl ester, favoring the (S)-enantiomer. histidine, but the molecular basis for disruption dif-
We synthesized phosphonate (RC,RPSP)-3, which, fers in each case showing that multiple mechanisms
upon reaction with CRL, mimics the transition state are required for high enantioselectivity.
for hydrolysis of (S)-1-methyl ester, the fast-reacting
enantiomer. An X-ray crystal structure of this com- Keywords: carboxylic acids; enantioselectivity;
plex shows a catalytically productive orientation with enzyme catalysis; hydrolases; molecular dynamics;
the phenyl ring in the hydrophobic tunnel of the phosphonates; protein structures; transition-state an-
lipase. Phe345 crowds the region near the substrate alogs; X-ray diffraction
Introduction
matory drugs and 2-aryloxypropionic acids, such as di-
chlorprop (aryloxy=2,4-dichlorophenoxy), which are
Lipases often accept unnatural ester substrates and herbicides, and can also reduce blood cholesterol
show high enantioselectivity and are therefore useful levels.[1,2]
in organic synthesis. One of the best-studied lipases
An empirical rule predicts the favored enantiomer
for its enantioselectivity toward chiral acids is lipase of chiral carboxylic acids for Candida rugosa-cata-
from Candida rugosa (CRL). This lipase shows high lyzed reactions[3] (Figure 1). The rule applies to car-
enantioselectivity toward many carboxylic acids with boxylic acids with a stereocenter at the a-position and
a stereocenter at the a-position. Examples include 2- relies on the sizes of the substituents at the stereocen-
arylpropionic acids such as naproxen [aryl=2-(6-me- ter. This rule suggests that CRL distinguishes enantio-
thoxy)naphthyl], which are non-steroidal anti-inflam- mers based on the size of the substituents.
Adv. Synth. Catal. 2011, 353, 2529 – 2544
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2529

Ian J. Colton et al.
FULL PAPERS
zyme composition, the purification procedure can also
change the enantioselectivity, but not the enantiopre-
ference, of CRL.[14,15,16] The working hypothesis is that
purification steps such as treatment with organic sol-
vents can change the conformation of the lipase, pos-
sibly from the closed to the open conformation.
Commercial samples of CRL show low enantiose-
lectivity (E=8) toward methyl 2-methoxy-2-phenyla-
cetate (1-methyl ester), favoring the (S)-enantiomer[4].
Purified CRL (A-form, which likely corresponds to
lip1) shows higher enantioselectivity: E=23 (S).[4]
This paper focuses on the molecular basis for this
enantioselectivity.
Figure 1. An empirical rule based on substituent size pre-
dicts the faster-reacting enantiomer in hydrolyses of esters
catalyzed by purified Candida rugosa lipase. a) Generic
structure of the acid predicted to form faster from the corre-
sponding ester. LA represents a large substituent, while MA
represents a medium substituent. b) An example of a car-
boxylic acid that fits this rule. Hydrolysis of racemic methyl
2-methoxy-2-phenylacetate favors the (S)-enantiomer shown
by a factor of 23 over the (R)-enantiomer.[4]
CRL-catalyzed hydrolysis of esters occurs in two
The molecular details of this enantioselectivity are steps: acylation of the active site serine followed by
still unclear. One cannot accurately predict the deacylation (Figure 2). The acylation step involves the
degree of enantioselectivity, nor can one predict first tetrahedral intermediate, Td1, while the deacyla-
which amino acid substitutions will increase enantio- tion step involves the second tetrahedral intermedi-
selectivity. An understanding of these details is essen- ate, Td2. The first irreversible step in hydrolysis is the
tial for rational design of enantioselective enzyme-cat- acylation step. (Although in principle, this step is re-
alyzed reactions. This understanding could also help versible, in practice the concentration of leaving
organic chemists design chemical catalysts with high group alcohol is low so it is effectively irreversible.)
enantioselectivity.
This acylation step is also likely the rate-determining
The compound 2-methoxy-2-phenylacetic acid (O- step because substrates with a better leaving group
methylmandelic acid, 1) is the focus of this work. react faster. Kinetic studies of another lipase showed
Enantiomerically pure 2-methoxy-2-phenylacetic acid that acylation was the rate-determining step.[17]
is a useful reagent for NMR analysis of the absolute
A tetrahedral intermediate exists for only a short
stereochemistry of amines and alcohols.[5] Lipase from time, but a stable phosphonate mimics this intermedi-
Aspergillus niger resolved the corresponding vinyl ate. We synthesized phosphonate inactivators that
ester of 1 by transesterification with methanol in iso- could react with the active site serine in CRL and
propyl ether with an enantioselectivity of 25,[6] but make a stable phosphonate. We solved the X-ray crys-
this lipase is not considered in this paper. This chiral tal structure of the CRL-phosphonate complex corre-
acid is similar in structure to the 2-arylpropionic acids sponding to the tetrahedral intermediate for the fast-
mentioned above and is a general example for the reacting enantiomer of 1-methyl ester. We also syn-
CRL-catalyzed hydrolysis of esters of chiral acids.
thesized the phosphonochloridate corresponding to
The genome of the yeast Candida rugosa contains the slow-reacting enantiomer, but it did not form a
at least seven lipase isoenzymes named lip1–7. The covalent complex with CRL. We investigated the ori-
isoenzymes differ in amino acid sequence (ꢀ80% entations of the slow-reacting enantiomer using mo-
amino acid sequence identity among lip1–5, ꢀ65% lecular modeling and identified the multiple molecu-
among 1–7), isoelectric point and degree of glycosyla- lar interactions required to prevent its reaction. This
tion.[7] Commercial samples of Candida rugosa lipase understanding is an essential first step to the rational
are non-recombinant enzymes and contain multiple design of higher enantioselectivity, but also shows the
isozymes. Gel electrophoresis (SDS-PAGE) of com- complexity and difficulty of the task.
mercial samples of CRL shows a major band and
minor band. The major band corresponds to lip1,
while the minor band is a mixture of isoenzymes lip2– Results
5. In addition, commercial samples of CRL contain a
small amount of contaminating protease.[8] The X-ray Synthesis of Transition State Analogs
crystal structures of three isoenzymes have been
solved: lip1,[9,10] lip2[11] and lip3.[12]
Racemic Inactivators
The isoenzymes differ in their enantioselectivity
toward acids and secondary alcohols, but show the Two racemic inactivators – the chloro and p-nitro-
same enantiopreference.[13,14] Isoenzyme lip1, the phenyl (PNP) derivatives, (ꢁ)-3 and (ꢁ)-4, were pre-
focus of this paper, shows the highest enantioselectivi- pared as diastereomeric mixtures (Figure 3). Benzal-
ty toward ibuprofen {2-[4-(2-methylpropyl)phenyl]- dehyde and dimethyl phosphite condensed over solid
propanoic acid}, which, among the compounds tested, KF to form dimethyl (1-hydroxyphenylmethyl)-
is closest in structure to compound 1. Besides isoen- phosphonate, (ꢁ)-2, according to a literature proce-
2530
asc.wiley-vch.de
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2529 – 2544

Molecular Basis of Chiral Acid Recognition by Candida rugosa Lipase
Figure 2. a) Mechanism of CRL-catalyzed hydrolysis of racemic methyl 2-methoxy-2-phenylacetate [(ꢁ)-1-methyl ester] in-
volves two tetrahedral intermediates and an acyl enzyme intermediate. The rate-determining step is the formation of the
acyl enzyme intermediate, which involves formation and collapse of the first tetrahedral intermediate. The transition state
for the reaction therefore resembles the first tetrahedral intermediate. For clarity, this Scheme omits the third member of
the catalytic triad, Glu341. b) A stable phosphonate mimics the first tetrahedral intermediate and the transition state for hy-
drolysis of (ꢁ)-1-methyl ester.
dure[18] in 88% yield. Methylation of (ꢁ)-2 using
methyl iodide and Ag2O afforded dimethyl [1-(me-
thoxy)phenylmethyl]phosphonate in 66% isolated
yield. This methylation was the most difficult reaction
to optimize in this synthesis. Initial trials using basic
conditions (Mel/K2CO3 in EtOAc or Mel/KOH in
DMSO) yielded an uncharacterized decomposition
product, likely due to reversion of (ꢁ)-2 back to start-
ing materials under basic conditions[19] and subse-
quent decomposition. Using the neutral silver oxide
to increase the electrophilicity of methyl iodide avoid-
ed these problems. A stoichiometric amount of
methyl iodide gave inconsistent yields for this methyl-
ation step, likely due to evaporation of methyl iodide,
but using methyl iodide as the solvent eliminated this
problem and was the optimum way to carry out this
step. Hydrolysis of dimethyl [1-(methoxy)phenylme-
thyl]phosphonate using aqueous base afforded the
monoacid in 85% yield. Monochlorination, using 1.5
equivalents of oxalyl chloride in refluxing dichlorome-
thane, yielded a 4:1 diastereomeric mixture of mono- PNP=p-nitrophenyl
Figure 3. Synthesis of racemic inactivators chloro derivative
(ꢁ)-3 and p-nitrophenyl derivative (ꢁ)-4. Only the chloro
derivative (ꢁ)-3 inactivated lipase from Candida rugosa.
Adv. Synth. Catal. 2011, 353, 2529 – 2544
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2531

Ian J. Colton et al.
FULL PAPERS
chlorophosphonates, (ꢁ)-3. Reaction of the mono- tate (R)-5 (65% recovery, 88% ee R). These enantio-
chlorophosphonates with p-nitrophenol and triethyl- meric purities correspond to an E of 32, significantly
ACHTUNGTRENNUNG
Inactivator (ꢁ)-3 inactivated lipase from Candida tiomeric purity, the acetate was resolved two more
rugosa, but inactivator (ꢁ)-4 did not. Incubation of a times with the same enzyme. Reincubation of the ace-
103-fold excess of (ꢁ)-3 dissolved in dichloromethane tate (68% ee R) with lipase for an additional 26.8%
with an aqueous solution of lipase from Candida conversion, then separation and reincubation for an
rugosa eliminated >99% of the hydrolytic activity additional 10.3% conversion yielded the remaining
within 8 min. In contrast, overnight incubation of a acetate in >98% ee R, 0.4 g, 14% overall yield (50%
104-fold excess of (ꢁ)-4 with CRL, eliminated <5% yield is the maximum in a resolution). This acetate
of the hydrolytic activity. Since the chlorophospho- was cleaved with triethylamine-methanol to yield (R)-
nate (ꢁ)-3 was a good inactivator, we prepared com- 2 for the next step in the synthesis. To prepare enan-
pound 3 with a single configuration at the carbon tiomerically pure (S)-2, we reacetylated enantiomeri-
center.
cally enriched (S)-2 to the acetate (S)-5 and carried
out a second hydrolysis catalyzed by the lipase until
46% conversion, which yielded (S)-2 in >98% ee,
0.11 g (4.4% overall yield).
Inactivator 3 Having a Single Configuration at the
Carbon Center
Methylation of (R)-2 used silver oxide and excess
methyl iodide as developed for the racemic alcohol,
The chloro derivative 3 with a single configuration at (ꢁ)-2. The overall yield for the hydrolysis of the ace-
the carbon center was prepared similarly, but with the tate and methylation was 50%. Hydrolysis of (R)-di-
addition of
a lipase-catalyzed kinetic resolution methyl [1-(methoxy)phenylmethyl]phosphonate to the
(Figure 4). Secondary alcohol (ꢁ)-2 was converted to monoacid used 1.5 equivalents of Nal in refluxing ace-
the acetate, (ꢁ)-5, which was resolved using lipase tone, followed by acidification to afford (R)-monoacid
from Rhizopus oryzae (Amano FAP 15) yielding (S)- (83% yield). These milder conditions for the hydroly-
2 (78% recovery, 68% ee) and recovered starting ace- sis, as compared to the NaOH used to hydrolyze the
racemate in Figure 3, minimized the possibility for
racemization at the resolved stereocenter. The enan-
tiomeric purity of 2 was measured after reacetylation
1
with acetic anhydride to the acetate 5, by H NMR in
the presence of a chiral shift reagent, EuACTHNURGTNEUNG(hfc)3, but
the enantiomeric purity of the subsequent compounds
could not be measured.
Finally, chlorination of the monoacid using 100
equivalents of of oxalyl chloride yielded a 4:1 mixture
of the two phosphonochloridates, (RC,RPSP)-3 as ob-
1
served by H and 31P NMR. Reaction of the monoacid
with less oxalyl chloride yielded a substantial amount
of pyrophosphonate side product, which did not inac-
tivate CRL.
An analogous route (not shown) formed phospho-
nate (SC,RPSP)-3, which mimics hydrolysis of the slow-
reacting (R)-enantiomer of 1-methyl ester.
X-Ray Crystal Structure
The covalent enzyme-inhibitor complex was prepared
by adding excess (RC,RPSP)-3 diastereomeric mixture
dissolved in 2-methyl-2,4-pentanediol to the purified
CRL solution. The inhibited protein crystallized
under conditions similar to those for the open form
CRL.[9] The crystal structure was solved by isomor-
phous replacement with the open form CRL struc-
ture. As expected from the synthesis, the configura-
tion at the carbon stereocenter was R (Figure 5). The
Figure 4. Synthesis of inactivator (RC,RPSP)-3, which, upon
reaction with lipase from Candida rugosa, mimics the transi-
tion state for the CRL-catalyzed hydrolysis of (S)-1 methyl
ester (fast-reacting enantiomer). A key step in the synthesis
of inactivator (RC,RPSP)-3 was a lipase-catalyzed resolution
of acetate (ꢁ)-5 by lipase from Rhizopus oryzae. An analo-
gous route (not shown) formed phosphonate (SC,RPSP)-3,
which mimics hydrolysis of the slow-reacting (R)-enantio-
mer of 1-methyl ester.
2532
asc.wiley-vch.de
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2529 – 2544

Molecular Basis of Chiral Acid Recognition by Candida rugosa Lipase
The overall structure is identical to the open form
of Candida rugosa lipase.[9] The lipase adopts the a/b
hydrolase fold. The protein scaffold is formed by a
central 11-stranded mixed b-sheet and an N-terminal
3-stranded b-sheet. The flap is in its open conforma-
tion and is stabilized by Glu66 which is preceded by a
proline at one hinge and by the Pro92 cis-trans iso-
merization at the other hinge. The X-ray structure of
the closed form of CRL[10a] was solved previously
using different crystallization conditions and without
a phosphonate inhibitor. The transition state analog
in the current structure is well defined in the electron
density. There are 3 sugar molecules and 175 water
molecules identified in the electron density. The over-
all geometry of the molecule is very good (Table 1)
with the final R/Rfree =17/21%, average B-factor of
22.6 ꢂ2 and good stereochemistry.
Table 1. Data collection and refinement statistics for x-ray
structure of the complex of inactivator (RC,RPSP)-2 with
Candida rugosa lipase, which mimics reaction of the fast-re-
acting enantiomer (S)-1.
Crystal parameters and diffraction data
X-ray source
CuKa
Space group
C2221
Unit cell dimensions (ꢂ)
a=65.0
b=97.5
c=176.1
Figure 5. Phosphonate transition state analog for hydrolysis
of (S)-2 in the active site of CRL. a) Schematic diagram. b)
The model of the phosphonate fits well in the observed elec-
tron density. The mesh represents the difference simulated
annealing omit map at the 3 sigma level. The inhibitor and
all protein atoms within a 3.5 ꢂ sphere from the inhibitor
were excluded from calculations. c) X-ray crystal structure.
The analog mimics a catalytically productive orientation be-
cause the phosphorus links covalently to the catalytic serine
Data collection statistics
Total number of accepted reflections
51,648
23,670
(2,716)
54.1–2.2
2.3–2.2
9.2 (23.0)
10.5 (3.1)
81.0 (62.5)
Total number of unique reflectionsa
Resolution range (ꢂ)
Highest shell (ꢂ)[a]
Rmerge (%)[b]
Mean I/sI
Completeness (%)
ꢂ
(Ser209) and the analog makes good hydrogen bonds (O N
distances <3.2 ꢂ) to the catalytic histidine (His449; shown
as sticks for clarity) and the oxyanion hole (Ala210,
Gly124). The carbon stereocenter lies at the entrance to a
tunnel with the phenyl group pointing inside the tunnel.
Residues Phe344, Phe345, and Phe415 (mostly hidden by
Phe344 and Phe345) block the left side of the tunnel en-
trance. A hydrogen substituent can fit here, but not an OMe
substituent. Residues Leu297 and Phe296 form the right
side of the tunnel entrance. Site-directed mutagenesis ex-
periments previously identified Phe344 and Phe345 as im-
portant for the enantioselectivity of CRL toward 2-arylpro-
pionic acids.
Refinement statistics
Protein molecules in asymmetric units
Number of water molecules
Number of sugar molecules
Number of Ca2+
1
175
3
2
Final R/Rfree factors[c]
Average B-factor (ꢂ2)
r.m.s deviations bond lengths (ꢂ)
r.m.s deviations bond angles (o)
17/21
22.6
0.006
1.3
Ramachandran statistics: favored (%), allowed 93.4, 5.1,
(%), outlier (%)
1.5
[a]
Data for the highest resolution shell are given in paren-
theses.
[b]
[c]
R
merge = jShkl Si jI
G
ACHTUNGTRENNUNG
where I(hkl) is the observed intensity and <ICAHTUNGTRENNNUG
U
configuration at phosphorus was also a single configu-
ration (SP) although the inhibitor was a mixture of
both configurations at phosphorus. Presumably, the
inhibitor with one phosphorus configuration reacted
much faster that the other leading to only the fast-re-
acting phosphorus configuration bound to the lipase.
the average intensity obtained from multiple observa-
tions of symmetry-related reflections after rejection.
R=Sj jFo jꢂjFcj j/SjFo j, where Fo and Fc are the ob-
served and calculated structure factors, respectively. Rfree
is calculated using 9.9% of the reflections randomly ex-
cluded from refinement.
Adv. Synth. Catal. 2011, 353, 2529 – 2544
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2533

Ian J. Colton et al.
FULL PAPERS
The crystal structure of the transition state analog Modeling to Find Possible Orientations of the Slow-
confirms that the stereocenter matches the fast-react- Reacting Enantiomer
ing enantiomer because of the good fit of this stereo-
isomer with the measured electron density (Figure 5). To identify the orientation of the slow enantiomer, we
The electron density was obtained by running a simu- used molecular modeling. There are four ways that
lated annealing refinement with the analog and the slow enantiomer can bind in the same place in
nearby protein atoms (within 3.5 ꢂ) omitted. The cal- CRL as the fast enantiomer. Three of these possibili-
culated unbiased difference map confirms the selected ties involve an exchange of two substituents at the
enantiomer.
stereocenter and one possibility involves an umbrella-
The carbon stereocenter of the inhibitor lies at the like inversion orientation[21] (Figure 6). An analogy to
mouth of a tunnel with the phenyl ring pointing these different possibilities is the two different ways
inside the hydrophobic tunnel (Figure 5). This tunnel that you can overlay your right and left hand,
accommodates the fatty acyl chains of other inhibi- Figure 6 (b). In one way the two palms face in the
tors. The methoxy substituent at the carbon stereo- same direction, but the finger do not align, which is
center points toward Phe296, while the hydrogen at similar to one of the exchange of substituents possibil-
the carbon stereocenter points toward Phe344.
ities. In the second way, the palms face in opposite di-
Similar crystals formed using the phosphonochlori- rections, but fingers line up correctly, which is similar
date mimicking the slow-reacting enantiomer showed to the umbrella-like inversion. High enantioselectivity
no inhibitor in the active site. Presumably the slow re- requires the lipase to prevent reaction of the slow
acting enantiomer fits so poorly that this inhibitor did enantiomer in any of these four orientations.
not react with the lipase.
To identify how CRL prevents hydrolysis of (R)-1
in all four possible orientations, we modeled tetrahe-
Figure 6. Possible orientations of the slow enantiomer of 1 in the active site of Candida rugosa lipase. a) Schematic of the
fast-reacting enantiomer of methyl 2-methoxy-2-phenylacetate [(S)-1, left] and three possible exchange-of-substituents orien-
tations for the slow enantiomer [(R)-1, right three structures]. The curves lines represent the active site; gray curves indicate
the two positions that have mismatched substituents for the slow enantiomer. The exclamation points mark points where
steric interactions disfavor the three orientations for the slow enantiomer. b) Schematic of the fast-reacting enantiomer (S,
left) and an umbrella-like inversion orientation of the slow enantiomer (R, right) in the same site. The mirror plane shown
relates the two substrate molecules. This reflection places the hydrogen substituent of the slow enantiomer in a new location
leaving the original site empty (gray curve). c) An analogy to these different possible orientations is the two ways to overlay
your right hand on top of your left hand. On the left panel, the fingers match with each other, but the palms face in opposite
directions: bottom hand palm faces down, while top hand palm faces up. On the right panel, the fingers do not overlay cor-
rectly, but both palms face down.
2534
asc.wiley-vch.de
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2529 – 2544

Molecular Basis of Chiral Acid Recognition by Candida rugosa Lipase
dral intermediates corresponding to these four possi- and collapse of a tetrahedral intermediate. The struc-
bilities. These tetrahedral intermediates were built tures of the transition states for formation and col-
using the crystal structure of the phosphonate analog lapse of the tetrahedral intermediate are similar to
for hydrolysis of (S)-1 as the model. The modeling the tetrahedral intermediate, so this modeling ap-
used the OPLS 2005 force field[24] for the lipase and proach models the reactivity of each case.
tetrahedral intermediate and a TIP3P model[25] for
Modeling correctly predicts that the tetrahedral in-
the water solvent. A molecular dynamics simulation termediate for hydrolysis of (S)-1-methyl ester adopts
generated multiple conformations of each possible a catalytically productive orientation (Table 2). The
orientation to mimic the multiple conformations pos- geometry of the 250 structures generated by the mo-
sible during the reaction.
lecular dynamic simulation were measured to identify
The tetrahedral intermediate must form four cata- hydrogen bonds between the lipase and the tetrahe-
lytically important hydrogen bonds with the lipase: dral intermediate. Hydrogen bonds were defined as
ꢂ
ꢂ
ꢂ ꢂ
from the two N H bonds of the oxyanion hole an O N distance of <3.55 ꢂ and an O H N angle
(Ala124, Gly210) to the oxyanion oxygen of the tetra- >1208 (1808 is ideal). These criteria are permissive to
ꢂ
hedral intermediate and from Ne H of the catalytic include the entire range of hydrogen bond distances
histidine to the alcohol oxygen of the tetrahedral in- observed in proteins.[23] The (S)-enantiomer formed
termediate and to Og of active site serine, see all four catalytically essential hydrogen bonds to the
Figure 3 above. Conformations containing all five cat- protein in 51% of the structures, which is consistent
alytically essential hydrogen bonds were considered with the experimental result that (S)-1-methyl ester is
catalytically productive, while those missing one or a good substrate. The structures of the modeled tetra-
more hydrogen bonds were considered non-produc- hedral intermediate were similar to the phosphonate
tive. The fraction of catalytically productive confor- X-ray crystal structure; Figure 7 shows a typical struc-
mations among all the conformations generated by ture.
molecular dynamics in each case indicates the likeli-
Modeling also correctly predicts that the slow enan-
hood that the reaction can occur. Since the enantiose- tiomer reacts very slowly because it only rarely
lectivity of CRL lip1 toward 1 is 23, we expect the adopts a catalytically productive orientation (Table 1).
fraction of catalytically productive conformations to Almost all orientations disrupt at least one of the two
be 23-fold lower for (R)-1 as compared to (S)-1.
hydrogen bonds to the catalytic histidine. The excep-
This modeling approach is similar to the near- tion is the (R)-exchange H/OMe orientation where
attack-complex modeling developed by Bruice and one of the 250 modeled conformations (0.4%)
co-workers.[22] A near attack complex is not an inter- showed intact catalytic hydrogen bonds. The rarity of
mediate, but a fleeting structure where the distances this catalytically productive conformation indicates
and angles are close to those needed for the reaction. that reaction of the (R)-enantiomer is much slower
Our approach differs by not modeling a near attack than for the (S)-enantiomer.
complex, but the tetrahedral intermediate. Lipase-cat-
The (R)-exchange H/Ph orientation disrupts three
alyzed hydrolysis of an ester involves the formation key hydrogen bonds in most conformations: two from
Table 2. Percentage of conformations that contain a hydrogen bond between the partners indicated.[a]
Oxyanion oxygen of tetrahedral
intermediate
His449-Ne-H
Conformation Ala124-N-H, % Gly210-N-H, % Ser209-Og, % Me-O, % All four productive Glu208-O
H-bonds, %
(non-
productive), %[b]
(S)
82 (100)
100 (100)
100 (100)
12 (100)
93 (89)
0.79 (0)
78 (88)
87 (0)
51
0
7.5 (59)
100 (100)
(R)-exchange
H/Ph
(R)-exchange
OMe/Ph
(R)-exchange
H/OMe
100 (100)
100 (100)
98 (100)
100 (100)
100 (100)
60 (1.6)
100 (99)
0.39 (0)
0.40 (2.8)
0.39 (99)
65 (0)
0
99 (98)
0.39
0
25 (0.40)
100 (100)
(R)-umbrella
100 (100)
[a]
ꢂ
Percentage of the structures where the N O distance is <3.55 ꢂ. The numbers in parentheses indicate the percentage of
the structures where the N H O angle is >1208. The data is in Figure S-1, Figure S-2, and Figure S-3 in the Supporting
Information.
ꢂ ꢂ
[b]
In some non-productive structures a catalytically non-productive hydrogen bond formed between the catalytic histidine
and the side chain of Glu208.
Adv. Synth. Catal. 2011, 353, 2529 – 2544
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2535

Ian J. Colton et al.
FULL PAPERS
the catalytic histidine to the catalytic serine Og and to and Phe345 restrict the space available for the phenyl
ꢂ
the substrateꢃs ester oxygen and one from Ala124 N
H to the oxyanion oxygen. The side chains of Phe415 moves the imidazole ring (Figure 7). The hydrogen
group, so it bumps into the catalytic histidine and
Figure 7. Representative structures from molecular dynamics modeling of tetrahedral intermediates for hydrolysis of (S)-
and (R)-1. For (S)-1 (gray), all four catalytically important hydrogen bonds are present (dotted lines). In all four orientations
of (R)-1 [green overlaid on the gray structure of (S)-1], one or both of the hydrogen bonds from the catalytic histidine to
either the catalytic serine Og or the substrateꢃs ester oxygen are missing.
2536
asc.wiley-vch.de
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2529 – 2544

Molecular Basis of Chiral Acid Recognition by Candida rugosa Lipase
bond donor atom, Ne moves by 3.3 ꢂ as compared to bond forms 31% of the time. This hydrogen bond is
ꢂ
the (S) enantiomer and thus breaks the two hydrogen non-productive because the N H cannot donate a
bonds between His-Ne the tetrahedral intermediate. proton to leaving group OMe because the pKa of an
ꢂ
The overall tetrahedral intermediate moves by 1.0 ꢂ amide N H (~17) is higher than the pKa of methanol
towards the readerꢃs left, which disrupts a key hydro- (15).
gen bond between the oxyanion and main-chain
The slow enantiomer most likely reacts via either
amide of Ala210. None of the conformations identi- the (R)-exchange H/OMe orientation or the (R)-um-
fied by molecular dynamics contain all four essential brella orientation. These orientations are closest to a
hydrogen bonds; thus, none were catalytically produc- catalytically productive orientation because they lack
tive.
only one hydrogen bond as compared to three for
The (R)-exchange OMe/Ph orientation disrupts the (R)-exchange H/Ph and two for (R)-exchange OMe/
two key hydrogen bonds between the catalytic histi- Ph. The other three orientations also make a non-pro-
dine and the catalytic serine Og and the substrateꢃs ductive hydrogen bond with the catalytic histidine all
ester oxygen. The initial model for this orientation the time, but this exchange H/OMe orientation makes
places the phenyl group point upward where it clashes non-productive hydrogen bonds only 56% of the
with Phe296. During geometry optimization, the sub- time. This non-productive hydrogen bond occupies
strates phenyl group moves, bumps into the catalytic the histidine and prevents reaction. We hypothesize
histidine and moves the imidazole ring (Figure 7). that protein engineering efforts to increase enantiose-
The hydrogen bond donor atom, Ne moves by 3.6 ꢂ lectivity should focus on further destabilizing the (R)-
as compared to the (S) enantiomer and thus breaks exchange H/OMe orientation and the (R)-umbrella
the two hydrogen bonds between His-Ne the tetrahe- orientation.
dral intermediate.
The modeling does not quantitatively match the ex-
The (R)-exchange H/OMe orientation disrupts only perimental results. We identified only one out of 1000
one hydrogen bond – between the catalytic histidine modeled conformations (0.1%) for the slow reacting
and the substrateꢃs ester oxygen. The methoxy group enantiomer that were catalytically productive. Since
at the stereocenter bumps into Phe344 causing the 51% of the conformations were catalytically produc-
entire tetrahedral intermediate to shift 0.7 ꢂ to the tive for the fast-reacting enantiomer and the slow-re-
readerꢃs right. The shift breaks a key hydrogen bond acting enantiomer reacts 23-fold slower, quantitative
between His449 and the ester oxygen. In one confor- agreement requires ~2% of the conformation for the
mation out of 250 (0.40%) all catalytically essential slow-reacting enantiomer to be catalytically produc-
hydrogen bonds are present.
tive. The reason for the lack of agreement is unknown
The (R)-umbrella orientation disrupts only one hy- and could be due to imperfect modeling or due to an
drogen bond – between the catalytic histidine and the underestimate of the enantioselectivity; for example,
catalytic serine Og. As the phenyl group moves to a if the enzyme contained a contaminating less enantio-
position similar to the (S)-enantiomer, the methoxy selective isozyme.
group at the stereocenter bumps into the catalytic his-
tidine and moves the imidazole ring (Figure 7). The
hydrogen bond donor atom, Ne moves by 3.0 ꢂ as Discussion
compared to the (S) enantiomer and thus breaks the
hydrogen bonds between His-Ne and the catalytic Orientation of the Fast-Reacting Enantiomer
serine Og.
In place of the disrupted hydrogen bonds between Previous computer modeling and extrapolation from
the catalytic histidine and either the catalytic serine X-ray crystal structures indicated that chiral acids
or substrate ester OMe, a new hydrogen bond forms with an a-stereocenter bind to CRL with the stereo-
between His-Ne and Glu-O (Table 2). This hydrogen center positioned at the mouth of the tunnel and the
bond is not catalytically productive because it does large substituent extending into the tunnel. The X-ray
not stabilize the tetrahedral intermediate. The fast-re- structure reported in this paper confirms this hypothe-
acting enantiomer (S)-1-methyl ester forms this hy- sis. The phenyl group lies inside the tunnel, while the
drogen bond only rarely, 7.5% of the time. In con- medium substituent (methoxy) points upward near
trast, the slow-reacting enantiomer orientations usual- Phe296 and Leu297 residues, while the hydrogen
ly make this non-productive hydrogen bond. Three points toward Phe344, Phe345 and Phe415. The bind-
orientations – exchange H/Ph, exchange OMe/Ph, and ing site for chiral secondary alcohols overlaps with
umbrella – make it nearly all the time. The (R)-ex- this region. The hydrophobic region consisting of
change H/OMe orientation make this hydrogen bond Phe296, Phe344, Phe345 and Leu297 also corresponds
only 25% of the time, but it also forms another non- to the ꢄlarge pocketꢃ for secondary alcohols.
productive hydrogen bond between the ester oxygen
and the main-chain amide of Gly123 – this hydrogen
Adv. Synth. Catal. 2011, 353, 2529 – 2544
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2537

Ian J. Colton et al.
FULL PAPERS
Orientation of the Slow-Reacting Enantiomer
esterification of 2-methyldecanoic acid from E=83
(90 mM n-heptanol) to E=37 (900 mM n-hepta-
Previous modeling and experiments with chiral acids nol).[30] The decreased enantioselectivity stems from
proposed differing orientations of the slow-reacting slower reaction of the fast enantiomer, but no change
enantiomer, likely because the researchers studied in the reaction rate of the slow enantiomer. To fit the
different chiral acids. Both the proposed orientations slow-reacting enantiomer, Holmquist and co-workers
are exchange-of-substituents conformations, while our first exchanged the locations of the hydrogen and
proposed umbrella-like inversion orientation is a new methyl group, but found that this orientation dis-
proposal.
placed the catalytic histidine and was therefore not
Molecular modeling of the CRL-catalyzed resolu- catalytically productive. Next, they exchanged the
tion 2-arylpropionic acids suggested that the slow methyl and n-alkyl groups. The methyl group pointed
enantiomer reacts via a hydrogen/medium (methyl) into the tunnel, while the alkyl group was near
exchange-of-substituents orientation.[24,25,26] Modeling Phe296 and extended toward the solvent. Enantiose-
naproxen (aryl=6-methoxy-2-naphthyl) in the active lective inhibition by the long-chain alcohols occurs
site of CRL placed the aryl group in the tunnel. For when the alcohol binds in the tunnel. This binding
the fast-reacting enantiomer, the methyl group at the does not affect the binding of the slow-reacting enan-
stereocenter pointed toward Phe296, as does the me- tiomer, but it blocks the fast-reacting enantiomer, re-
thoxy group in the phosphonate structure in this sulting in lower enantioselectivity.
paper. For the slow reacting enantiomer, the hydro-
Further support for this exchange of medium and
gen and methyl group at the stereocenter exchange large group locations came from a reversal of enantio-
positions (Figure 8). This is the same conclusion that selectivity for substrates that do not fit into the
we came to for 1-methyl ester, whose structure is clos- tunnel.[31] The CRL-catalyzed hydrolysis of ethyl 2-
est to the 2-arylpropionic acids. Modeling of 2-(4-sub- methyl-6-(2-thienyl)hexanoate favored the (R)-enan-
stituted phenoxy)propionic acid also suggested that tiomer, in contrast to (S)-2-methyldecanoate. The 2-
the methyl and hydrogen exchange locations when thienyl group made the alkyl chain too large to fit in
the slow enantiomer reacts.[27]
the tunnel, so it binds near Phe 296 and extends
The enantioselectivity of CRL toward naproxen in- toward the solvent. This reversal supports the notion
creases at higher temperatures indicating that entrop- that the methyl/n-alkyl exchange is a catalytically pro-
ic effects favor the fast reacting enantiomer.[28] The ductive orientation and can account for the reaction
fast enantiomer may be more flexible in the active of the slow enantiomer of 2-methyldecanoic acid.
site, while the slow enantiomer can only react via a (The experiments in refs.[34–37] mentioned above used
rarely reached conformation.
crude CRL so minor isozymes may contribute to the
In contrast, the slow enantiomer of 2-methylalkano- results. The major isozyme lip1 is most likely the
ic acids, where the large group is n-alkyl, likely reacts main contributor.)
in reacts via a medium (methyl)/large (n-alkyl) ex-
Increasing the enantioselectivity of CRL toward
change-of-substituents orientation.[29] The experimen- chiral acids is difficult for two reasons. First, different
tal evidence is that adding long-chain alcohols de- orientations of the slow enantiomer are possible.
creases the enantioselectivity of the CRL-catalyzed High enantioselectivity requires preventing reaction
Figure 8. Different proposed orientations of the slow enantiomer in CRL-catalyzed resolutions of chiral acids. a) For 2-aryl-
propionic acids, the slow enantiomer likely reacts via a medium (methyl)/hydrogen exchange. b) For 2-methylalkanoic acids,
where the large group (n-alkyl) is more flexible, the slow enantiomer likely reacts via a medium (methyl)/large (n-alkyl) ex-
change.
2538
asc.wiley-vch.de
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2529 – 2544

Molecular Basis of Chiral Acid Recognition by Candida rugosa Lipase
via all of them, which is complex since changes that ibuprofen.[14] Since these isoenzymes differ at position
hinder one orientation may help another orientation. 344 (Phe in lip1 versus Ile in lip3), the similar enantio-
Second, exchanging amino acids near the substrate is selectivity indicates that this position is not essential
a gigantic change on a molecular scale. Changing one for enantioselectivity. In contrast, lip2, which has a
amino acid, for example, Phe to Leu, removes several valine at 296 instead of phenylalanine in lip1 and lip3,
atoms; assuming no addition rearrangements, the side has lower enantioselectivity toward ibuprofen (E=
chain is now at least 2 ꢂ further from the substrate. 15). Isoenzyme lip2 also has a leucine at 344, but this
This is a large distance on a molecular scale. Contrast is unlikely to affect enantioselectivity because valine
this change to the ability to increase the cylinder size or isoleucine at this position did not affect enantiose-
in an engine by a millimeter, which is a much more lectivity.
subtle change. The alternative in protein engineering
Mutations in the lid region, which is far from the
is to change not the amino acids directly in contact substrate binding site, had minor effects on enantiose-
with the substrate, but the next sphere so that the lectivity, most likely by altering the proportion of
large changes in structure are more distributed and open and closed forms.[32] Researchers replaced the
effect on the substrate-binding site is more subtle.
lid region (residues 66–93) in lip1 with the corre-
Site-directed mutagenesis experiments and a com- sponding sequence in lip3 to create lip1lid3. This re-
parison of isoenzymes identified Phe345 and Phe296, placement introduced six changes in the amino acids
but not Phe344, as important to the high enantioselec- sequence at positions 69, 74, 76, 88, 91, and 93. The
tivity of CRL (lip1) toward 2-arylpropionic acids, but changed amino acids do not contact the substrate and
did not identify substitutions to increase the enantio- the closest one lies >15 ꢂ from the phosphonate.
selectvity of CRL (lip1) (Table 3). Site-directed muta- Under some conditions, the enantioselectivity of
genesis indicates that Phe345 is essential for high lip1lid3 was slightly lower than the enantioselectivity
enantioselectivity toward naproxen, a 2-arylpropionic of lip1 toward the 2-chloroethyl ester of 2-hydroxy-
acid, but that Phe344 is not essential. Substitution of caproic acid. The reaction was a transesterification
Phe345 with valine decreased the enantioselectivity of with methanol in isooctane. The enantioselectivities
CRL (lip1 isozyme, same as the one in this paper) were similar (E=10 for lip1, 8 for lip1lid3) with a
from 32 to 2.6.[30] Substitution of Phe344 with valine water activity of 0.06, but differed slightly (E=20 for
had no significant effect on enantioselectivity. The lip1, 13 for lip1lid3) at a water activity of 0.53. The
enantioselectivity changes with ketoprofen, another 2- authors suggested that catalysis occurs in the closed
arylpropionic acid, were similar.[30] Comparison of the form of the lipase with low enantioselectivity and in
enantioselectivity of CRL isoenzymes[7] also indicates the open form of the lipase with higher enantioselec-
that Phe344 is not essential for enantioselectivity and tivity. The substitutions on the lid alter the proportion
further identifies Phe296 as essential for high enantio- of open and closed forms. These experiments do not
selectivity toward ibuprofen, also a 2-arylpropionic help much in identifying key residues.
acid. Three lipase isoenzymes (lip1–3) have 80% of
Mutations within the tunnel of a related lipase,
the amino acids identical, but two amino acids differ lipase from Geotrichum candidum, which also con-
in the tunnel entrance region where the a-steroecen- tains a tunnel, could increase the enantioselectivity
ter of a carboxylic acid binds (Table 3). Isoenzymes toward 2-methylalkanoic acids from E=32 to 54, a
lip1 and lip3 showed high enantioselectivity toward moderate increase.[33] These amino acids were far
from the stereocenter and the molecular basis for this
change is unknown.
Table 3. Effect of amino acid substitutions near the mouth
of the tunnel on the enantioselectivity of CRL toward 2-ar-
ylpropionic acids. Substitutions in bold decrease the enantio-
selectivity strongly, while the other substitutions have little
effect.
Conclusions
In spite of these extensive and detailed studies of the
Isozyme
lip1
296 344 345
E
Refs.
enantioselectivity of CRL toward chiral a-substituted
carboxylic acids, researchers still cannot predict which
amino acid substitutions will increase enantioselectivi-
ty. The problem is difficult because all possible orien-
tations of the slow enantiomer must be considered. In
particular, the umbrella-like inversion orientation has
not been previously considered, but the computer
modeling in this paper suggests that the slow enantio-
mer can react via this orientation. The molecular dy-
namics showed that all four possible orientations of
the slow enantiomer were stable conformations and
[14,30]
Phe Phe Phe >50; ibupro-
fen
32; naproxen
[30]
[30]
[30]
lip1 - Phe344Val
lip1 - Phe345Val
lip1 -
Phe344,345Val
lip2
Phe Val Phe 29; naproxen
Phe Phe Val 2.6; naproxen
Phe Val Val 7.6; naproxen
[14]
[14]
Val Leu Phe 15; ibuprofen
Phe Ile Phe >50; ibupro-
fen
lip3
Adv. Synth. Catal. 2011, 353, 2529 – 2544
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2539

Ian J. Colton et al.
FULL PAPERS
identified the ones most likely to react, thereby lower- (ꢁ)-Methyl [1-(Methoxy)phenylmethyl]phosphonic
ing the enantioselectivity. Extending this approach to Acid
test possible amino acid substitutions may enable ra-
tional design of more enantioselective lipases.
The above (ꢁ)-dimethyl [1-(methoxy)phenylmethyl]phosph-
onate (0.60 g, 2.6 mmol) was hydrolyzed to the mono acid
by refluxing in aqueous sodium hydroxide (2.7M) for 5 h.
The solution was acidified to pH 1 with 1N HCI and then
extracted with CH2Cl2 (2ꢅ50 mL) and ether (2ꢅ50 mL).
Experimental Section
Further extractions with CH2Cl2 (2ꢅ50 mL) and EtOAc (1ꢅ
50 mL) did not remove the majority of the product, so the
water was evaporated off under reduced pressure and the
residue was extracted with CH2Cl2 (3ꢅ50 mL). The com-
bined organic extracts were dried with magnesium sulfate,
filtered and evaporated under reduced pressure affording
General
Chemicals and lipase from Candida rugosa (L-1754) was
purchased from Sigma Aldrich. FAP 15 lipase (Rhizopus
oryzae) was purchased from Amano International Enzyme
1
the mono acid; yield. 0.46 g 82.2 mmol, 85%): 1H NMR.
Co. (Troy, VA). H NMR spectra were recorded in CHCl3 at
2
200 MHz with CHCl3 as an internal standard (d=7.26).
13C NMR spectra were recorded in CHCl3 at 50 MHz with
CHCl3 as an internal standard (d=66.5) and were proton
decoupled. 31P NMR spectra were recorded in CHCl3 at
121 MHz with 85% phosphoric acid (0 ppm) as an external
standard.
d09.91 (br s, 1H), 7.40–7.31 (m, 5H), 4.42 (d, 1H, JH;P
=
15.8 Hz), 3.60 (d, 3H, 3JH;P =10.6 Hz), 3.38 (s, 3H);
13C NMR: d=134.0, 128.2, 127.9 (d, J=5.2 Hz), 79.8 (d, JC,P
1
171.0 Hz), 58.6 (d, 3JC,P =14.2 Hz), 53.1 (d, 2JC,P =6.7 Hz);
31P NMR: d=23.3 (s); MS (CI/NH3): m/z (rel. intensity)=
217 (16, M+H+), 121 (100).
(ꢁ)-Methyl [1-(Methoxy)phenylmethyl]phosphono-
chloridate [(ꢁ)-3]
(ꢁ)-Dimethyl (1-Hydroxyphenylmethyl)phosphonate
[(ꢁ)-2]
Oxalyl chloride (0.80 g, 0.55 mL, 6.3 mmol) was added to a
solution of (ꢁ)-methyl [1-(methoxy)phenylmethyllphos-
phonic acid (0.0136 g, 0.063 mmol) in dichloromethane
(13 mL) and the mixture was refluxed for 18.5 h under N2.
Evaporation of the solvent under reduced pressure yielded
(ꢁ)-3 as a mixture of diastereomers of the phosphonochlori-
date (yield: 70%) and pyrophosphonate derivatives [yield:
Following a literature procedure,[19] benzaldehyde (0.53 g,
0.51 mL, 5 mmol) and dimethyl phosphite (0.55 g, 0.46 mL,
5 mmol) were consecutively added to anhydrous potassium
fluoride (25.8 mmol, 1.50 g). The reaction mixture was man-
ually stirred until it thickened and then solidified within
5 min. Twenty-five mL of CH2Cl2 were added and the solid
was extracted for 5 min. The suspension was filtered and the
filtrate evaporated under vacuum to afford (ꢁ)-2 as a white
solid; yield: 0.95 g (4.39 mmol, 88%). 1H NMR: d=7.51–
7.27 (m, 5H), 5.05 (d, JH,P =11. 1 Hz), 4.68 (br s, 1H), 3.68
30%, 31P NMR: d=13.3 (app t)] as determined by 31P NMR.
2
1H NMR: d=7.51–7.39 (in, 5H), 4.74 (d, 1H, JH,P
=
=
3
11.4 Hz), 3.87 (d, 3H, 3JH,P =12.8 Hz), 3.86 (d, 3H, JH,P
12.8 Hz), 3.49 (s, 3H); 31P NMR: d=36.7 (s), 36.6 (s).
3
3
(d, 3H, JH,P =10.5 Hz), 3.66 (d, 3H, JH,P =10.3 Hz) [lit.[18]
:
1H NMR (CDCl3, 400 MHz): d=7.50–7.29 (m, 5H, C6H5),
5.05 (d, 1H, J=10.8 Hz, PhCH), 4.11 (br s, 1H, OH), 3.70,
(ꢁ)-Methyl p-
Nitrophenyl[1-ACTHUNRTGNEUNG(methoxy)phenylmethyl]-
3.67 (2ꢅd, 2ꢅ3H, J=10.3 Hz, PACHTNUGRTNEUNG
(OMe)2)]; 13C NMR: d=
phosphonate [(ꢁ)-4]
136.5, 128.3 (d, J=2.0 Hz), 128.1 (d, J=3.1 Hz), 127.0 (d,
J=6.0 Hz), 70.5 (d, 1JC,P =160 Hz), 53.9 (d, 2JC,P =7.1 Hz),
Oxalyl chloride (78 mL, 0.113 g, 0.89 mmol) was added to a
solution of (ꢁ)-methyl [1-(methoxy)phenylmethyl]phos-
phonic acid (0.128 g, 0.59 mmol) in dichloromethane (7 mL)
and the mixture was refluxed for 14 h under N2. The solvent
and excess oxalyl chloride were removed under reduced
pressure and the chlorophosphonate was redissolved in di-
chloromethane (7 mL). p-Nitrophenol (0.092 g, 0.65 mmol)
was added followed by triethylamine (91 mL, 0.066 g,
0.65 mmol) and the mixture was stirred overnight under N2.
The mixture was purified by column chromatography using
20:1 CH2Cl2/EtOAc (Rf =0.17) to afford a 1:1 mixture of
diastereomers (ꢁ)-4; overall yield: 0.040 g (1.2 mmol, 20%).
2
53.6 (d, JC,P 7.3 Hz); 31P NMR: d=24.3 (s).
(ꢁ)-Dimethyl [1-(Methoxy)phenylmethyl]-
phosphonate
(ꢁ)-2 (1 g, 4.6 mmol), silver(I) oxide (2.14 g, 9.2 mmol),
acetic acid (0.53 mL, 0.55 g, 9.2 mmol) and methyl iodide
(15 mL, 241 mmol) were combined and stirred in the dark
for 22 h. The reaction mixture was diluted with CH2Cl2, fil-
tered, and washed with 50 mL water. The CH2Cl2 layer was
separated, dried and filtered. Purification by flash chroma-
tography using 5:1 CH2Cl2/EtOAc (Rf =0.32) afforded the
3
1H NMR: d=8.17 (d, 2H, J=9.2 Hz), 7.52–7.34 (m, 5H),
3
2
7.24 (d, 2H, J=9.4 Hz), 4.74 (d, 1H, JH,P =15.4 Hz), 4.72
2
3
1
(d, 1H, JH,P =15.0 Hz), 3.84 (d, 3H, JH,P =11.0 Hz), 3.77 (d,
product; yield: 0.70 g (3.0 mmol, 66%). H NMR: d=7.46–
3H, JH,P =11.0 Hz), 3.43 (s, 3H); 31P NMR: d=17.8 (s); MS
3
7.36 (m, 5H), 4.56 (d, 1H, 2JH,P =15.8 Hz), 3.72 (d, 3H,
3JH,P =9.0 Hz), 3.67 (d, 3H, 3JH,P =8.8 Hz), 3.40 (s, 3H);
13C NMR: d=133.9, 128.4–128.2 (m), 127.7 (d, J=6.0 Hz),
80.0 (d, 1JC,P =168.9 Hz), 58.4 (d, 3JC,P =14.8 Hz), 53.5 (d,
(CI/NH3): m/z (rel. intensity)=338 (4.6, M+H+), 121 (100).
2
2JC,P =7.0 Hz), 53.3 (d, JC,P =6.9 Hz); 31P NMR: d=21.9 (s);
MS (CI/NH3): m/z (rel. intensity)=231 (27, M+H+), 121
(100).
2540
asc.wiley-vch.de
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2529 – 2544

Molecular Basis of Chiral Acid Recognition by Candida rugosa Lipase
filtered through a pad of diatomaceous earth. The filtrate
(ꢁ)-Dimethyl [1-(Acetyloxy)phenylmethyl]-
phosphonate [(ꢁ)-5]
was extracted with CH2Cl2 (3ꢅ100 mL). A small portion of
emulsion which remained throughout the extractions was
separately filtered through a small pad of diatomaceous
earth and washed with CH2Cl2. The filtrate of this filtered
emulsion was combined with the original CH2Cl2 extracts.
The starting material was separated from the product alco-
hol by column chromatography using 5:1 CH2Cl2/EtOAc in
order to elute the remaining substrate, (S)-5 (0.233 g,
0.90 mmol, 88% recovery), then 100% EtOAc to elute the
product alcohol, (S)-2 [0.112 g, 0.52 mmol, 59% recovery,
(ꢁ)-2 (0.200 g, 0.92 mmol), K2CO3 (0.26 g, 1.9 mmol), N,N-
dimethylaminopyridine (6 mg, 0.05 mmol) and acetic anhy-
dride (0.130 mL, 1.4 mmol) were added to ethyl acetate
(20 mL) and the mixture was stirred at room temperature
overnight. The mixture was filtered to remove K2CO3, and
the filtrate was washed with water (1ꢅ20 mL). The water
layer was extracted with EtOAc (2ꢅ25 mL), and the com-
bined organic extracts were dried with magnesium sulfate.
Evaporation of the solvent under vacuum afforded a clear
>98% ee by EuACHTNUGTRENUNG(hfc)3 chiral shift reagent following reacety-
lation].
1
oil; yield: 0. 19 g (0.75 mmol, 81%). H NMR: d=7.51–7.33
2
3
(m, 5H), 6.17 (d, 1H, JH,P =13.5 Hz), 3.73 (d, 3H, JH,P
=
:
To increase the enantiomeric purity of the acetate, (R)-5,
a sample (1.00 g, 3.9 mmol, 68% ee) was resolved a second
time. The sample was mixed with 3/1 hexanes/tert-butyl
methyl ether (20 mL) and phosphate buffer (50 mM,
pH 7.00, 130 mL). Lipase FAP 15 (505 mg dissolved in phos-
phate buffer, 50 mM, pH 7.00, 50 mL) and adjusted to
pH 7.00, was added and the pH was maintained at 7.00 with
0.50N NaOH by automatic titration using a pH stat. The re-
action was stopped after 16.5% conversion by adjusting the
reaction mixture to pH 4 using 1N HCl. The top organic
layer was separated. The aqueous layer was filtered through
a pad of diatomaceous earth and extracted with CH2Cl2 (3ꢅ
100 mL). The original organic layer from the enzyme reac-
tion and the combined CH2Cl2 extracts were filtered through
a fresh pad of diatomaceous earth and dried with MgSO4.
The acetate was purified by flash chromatography using 5:1
CH2Cl2/EtOAc to afford enantiomerically enriched (R)-5
[0.75 g, 2.9 mmol, 90% recovery based on the pH stat per-
10.7 Hz), 3.65 (d, 3H, 3JH,P =10.6 Hz), 2.19 (s, 3H) [lit.[20]
1H NMR (CDCl3, 400 MHz): d=7.50–7.32 (m, 5H, C6H5),
6.17 (d, 1H, J=13.3 Hz, PhCH), 3.72, 3.65 (2ꢅd, 2ꢅ3H, J=
10.3 Hz, PACHTUNGTRENNUNG(OMe)2), 2.18 (3H, s, MeCO).]
Kinetic Resolution of (ꢁ)-Dimethyl [1-(Acetyloxy)-
phenylmethyl]phosphonate [(ꢁ)-5]
(ꢁ)-5 (2.8 g, 10.8 mmol) was added to an Erlenmeyer flask
(500 mL) followed by a 3/1 mixture of hexanes/tert-butyl
methyl ether (20 mL) and phosphate buffer (10 mM,
pH 7.00, 100 mL). Lipase F-AP 15 (509 mg), dissolved in
phosphate buffer (50 mM, pH 7.00, 60 mL) and adjusted to
pH 7.00, was added to the phosphonate mixture. The pH
was maintained at 7.00 with 0.50N NaOH by automatic ti-
tration using a pH stat. The reaction was stopped at 43.8%
conversion by acidifying the reaction mixture to pH 4 with
1N HCl. The mixture was extracted with CH2Cl2 (1ꢅ
100 mL), EtOAc (1ꢅ50 mL) and again with CH2Cl2 (1ꢅ
100 mL). The combined organic extracts were filtered
through a pad of diatomaceous earth. The aqueous layer
was also filtered through a pad of diatomaceous earth and
the filtrate was further extracted with CH2Cl2 (1ꢅ100 mL).
The organic extracts were combined, dried with MgSO4, fil-
cent conversion, 92.7% ee by EuACHTNUGRTENUNG(hfc)3 chiral shift reagent].
A sample of this material (0.5 g, 1.9 mmol) was hydrolyzed
for an additional 10.3% conversion in the same manner to
afford (R)-5 [0.40 g, 1.6 mmol,>98% ee by EuACTHNURGTNEUNG(hfc)3 chiral
shift reagent].
tered and the solvent evaporated under reduced pressure. (R)-Dimethyl [1-(Methoxy)phenylmethyl]-
The remaining starting material acetate and product alcohol
were separated by flash chromatography using 5:1 CH2Cl2/
EtOAc to elute the acetate (R)-5 [yield: 1.22 g, 4.7 mmol,
phosphonate
(R)-5 (0.34 g, 1.33 mmol) was reacted with dry MeOH
(15 mL) and freshly distilled triethylamine (2 mL) for 18.5 h
at room temperature. The solvent and amine were removed
under reduced pressure and finally under high vacuum. To
this crude product was added Ag2O (0.62 g, 2.7 mmol), Mel
(15 mL) and acetic acid (0. 160 g, 0. 152 mL, 2.7 mmol) and
the mixture was stirred in the dark overnight. The methyl
iodide was then evaporated off with a stream of air. The
product was redissolved in CH2Cl2, filtered, concentrated
under reduced pressure, and purified by flash chromatogra-
phy using 5:1 CH2Cl2/EtOAc to afford (R)-2; overall yield:
78% recovery, 68% ee by Eu
ACHTUNGERTN(NUNG hfc)3 chiral shift reagent] and
then 100% EtOAc to elute the product alcohol (S)-2 [yield:
0.66 g, 3.1 mmol, 65% recovery, 88% ee by EuACTHNURGTNEUNG(hfc)3 chiral
shift reagent following reacetylation].
To increase the enantiomeric purity of the product alco-
hol, (S)-2, a sample was resolved a second time. (S)-2
(0.504 g, 2.3 mmol, 88% ee) was reacetylated with acetic an-
hydride (0.357 g, 0.330 mL, 3.5 mmol) and K2CO3 (0.650 g,
4.7 mmol) using N,N-4-dimethylaminopyridine (0.0156 g) as
catalyst in EtOAc (20 mL). The reaction mixture was fil-
tered, washed with dd H2O (50 mL), dried with MgSO4 and
the solvent evaporated under reduced pressure to afford
(S)-5; yield: 0.54 g (2.1 mmol, 90%). A sample of (S)-5
(0.49 g, 1.9 mmol, 88% ee) was dissolved with 3/1 hexanes/
tert-butyl methyl ether (10 mL) and added to phosphate
buffer (50 mM, pH 7.0, 100 mL). Lipase FAP-15 [220 mg dis-
solved in phosphate buffer (50 mM, pH 7.00, 50 mL)] was
added and the pH was maintained at 7.00 with 0.50N
NaOH by automatic titration using a pH stat. The reaction
was stopped at 46% conversion by acidifying the reaction
mixture to pH 3.9 with 1N HCl. The reaction mixture was
1
0.157 g (0.68 mmol, 50%). The H NMR spectrum was iden-
tical to that of the racemic standard.
AHCTNUGERT(GNNUN RC,RPSP)-Methyl [1-(Methoxy)phenylmethyl]-
phosphonic Acid
The above (R)-dimethyl [1-(methoxy)phenylmethyl]phos-
AHCTUNGTREGpNNUN honate (0.047 g, 0.21 mmol) was refluxed in freshly distilled
acetone with NaI (0.046 g, 0.31 mmol) for 22 h. The fine
product salt was filtered and rinsed with absolute ethyl
ether. The salt was dissolved in water, acidified to pH 1 and
Adv. Synth. Catal. 2011, 353, 2529 – 2544
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2541

Ian J. Colton et al.
FULL PAPERS
the water was evaporated off under reduced pressure. The
residue was extracted with dichloromethane (25 mL), dried
with MgSO4, filtered and the solvent removed under re-
duced pressure to afford the monoacid; yield: 0.036 g
(0.17 mmol, 81.3%). The H NMR spectrum was identical to
that of the racemic standard.
X-ray Crystal Structure
The covalent enzyme-inhibitor complex was prepared by
adding the inhibitor dissolved in 2-methyl-2,4-pentanediol
(MPD) to the protein solution. The inhibited protein was
crystallized under conditions similar to those used for native
protein.[9] The crystals are isomorphous with the native crys-
tals of CRL in the open form, Table 1. Diffraction data were
collected at 238C on an R-AXIS II image plate area detec-
tor and all refinements were done with CNS[34] and all mod-
eling was done using O[35] and COOT.[36] Coordinates were
deposited with the Protein Data Bank (www.pdb.org) and
have the code 3RAR.
1
ACHTUNGTRENNUNG(RC,RPSP)-Methyl [1-(Methoxy)phenylmethyl]-
phosphonochloridate [(RC,RPSP)-3]
Oxalyl chloride (1.59 g, 1.09 mL, 12.5 mmol) was added to a
solution of (R)-methyl [1-(methoxy)phenylmethyl]phos-
phonic acid (0.027 g, 0.125 mmol) in dichloromethane
(15 mL) and the mixture was refluxed for 18.5 h under N2.
Evaporation of the solvent under reduced pressure yielded
(RC,RPSP)-3 as a mixture of diastereomers according to
31P NMR. The 1H NMR spectrum was identical to that of
the racemic standard.
Molecular Modeling
Hydrogen atoms were added to the CRL-phosphonate X-
ray crystal structure using the software Maestro (Schrçding-
er, Portland OR; www.schrodinger.com) to correspond to a
pH of 7.0, except for the catalytic histidine, which was pro-
tonated. The sugars attached to CRL and the bound calcium
ions in the crystal structures were included in the simulation.
Water molecules far from heteroatoms (>5 ꢂ) were re-
moved. The geometry was optimized using the program
Macromodel v. 9.0 (Schrçdinger) using the OPLS 2005 force
field[37] until an rmsd (root mean square deviation) of
<0.3 ꢂ was reached. Atomic partial charges were calculated
using the bond increment method.[38] To model the tetrahe-
dral intermediate for hydrolysis of (S)-1, the phosphorus
atom was replaced by a carbon atom, the oxygen corre-
sponding to the oxyanion was assigned a formal negative
charge, and the geometry of the structure was optimized to
an rmsd of <0.05 ꢂ. To model the four different conforma-
tions of the tetrahedral intermediate for hydrolysis of (R)-1,
the substituents at the stereocenter were exchanged as indi-
cated in Figure 1 and the umbrella structure was generated
by moving the hydrogen atom to form the (R) enantiomer
without moving the other functional groups. Each conforma-
tion was modeled independently and was geometry opti-
mized to an rmsd of <0.05 ꢂ. Molecular dynamics simula-
tions used the software Desmond (D.E. Shaw Research,
New York, NY). The protein surrounded by 33885 water
molecules (TIP3P model[39]) to create a box sized 109 ꢂꢅ
109 ꢂꢅ95 ꢂ. Seventeen sodium ions were added to neutral-
ize the negative charge of the protein. The charge of ꢂ17
was verified using PROPKA 3.1,[40] which predicted a ꢂ17.5
charge at pH 7. This negative charge is about twice a large
as that for a typical protein because CRL is about twice the
size as a typical protein: 62 kDa versus 30 kDa for a typical
protein. Sixty-two additional sodium and chloride ions were
added to correspond to an ionic strength of 0.10M. The
system was equilibrated to 300 K and 1 atm pressure by
using Berendsen thermostat and barstat method[41] in six
separate stages. First, the solvent only was minimized; next
the solvent and protein were minimized. Stage 3 was a short
MD simulation (12 ps, velocity resampling every 1 ps) at
constant volume, moles and temperature (10 K), while stage
4 was a short MD simulation (12 ps, velocity resampling
every 1 ps) at constant pressure (1 atm), moles and tempera-
ture (10 K). Stage 5 was a slightly longer MD simulation
(24 ps) where the temperature is increased to 300 K and the
non-hydrogen atoms of the protein were fixed, while stage 6
was an MD simulation (24 ps) at 300 K, 1 atm with all atoms
(S)-Dimethyl [l-(Methoxy)phenylmethyl]phosphonate
To (S)-2 (0.108 g, 0.50 mmol,>98% ee) were added Ag2O
(0.232 g, 1.0 mmol), MeI (10 mL) and acetic acid (0.060 g,
0.057 mL, 1.0 mmol). The mixture was stirred in the dark
for 24 h. The MeI was then evaporated off with a stream of
air. The product was redissolved in CH2Cl2, filtered and con-
centrated under reduced pressure. The product was purified
by flash chromatography using 5/1 CH2Cl2/EtOAc to afford
(S)-dimethyl [l-(methoxy)phenylmethyl]phosphonate; yield:
0.058 g (0.25 mmol, 50%). This compound had an identical
1H NMR spectrum to the racemic standard.
ACHTUNGTRENNUNG(SC,RP,SP)-Methyl [1-(methoxy)phenylmethyl]phosphonic
acid was made starting with (S)-dimethyl [1-(methoxy)phe-
nylmethyl]phosphonate and subsequently converted to
(SC,RP,SP)-methyl
[1-(methoxy)phenylmethyl]phosphono-
chloridate using the same reaction conditions as described
above to prpeare the corresponding diastereomers that
mimic the slow-reacting enantiomer of 1.
Inactivation of Purified CRL using (ꢁ)-3
An aqueous solution of purified CRL (0.55 mg, 0.0092 mmol
in MES buffer, 10 mM, pH 6, 1 mL) was added to a di-
chloromethane solution of freshly synthesized chlorophosph-
onate, (ꢁ)-3 (9.2 mmol in 1 mL) and vigorously stirred for
eight minutes. The aqueous phase was removed, centrifuged
for 1 min and tested for activity using the enzyme assay de-
scribed below.
Enzyme Assay
An aliquot (5 mL) of enzyme solution followed by an aliquot
of p-nitrophenyl acetate (5 mL of a 50 mM solution in aceto-
nitrile) were added to phosphate buffer (1.0 mL, pH 7.5,
10 mM) at 258C. The initial rate of formation of p-nitrophe-
nolate was monitored at 404 nm for 30 s. An extinction coef-
ficient of 11600Mꢂl cmꢂl, which accounts for the incomplete
ionization of p-nitrophenolate at pH 7.5 was used to calcu-
late activity. Crude CRL showed 30 U/g solid (1 U/mg pro-
tein) with this assay. U=1 mmol of ester hydrolyzed per
minute.
2542
asc.wiley-vch.de
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2529 – 2544

Molecular Basis of Chiral Acid Recognition by Candida rugosa Lipase
free to move. The molecular dynamics simulation was for
1.2 ns with a step time of 1 fs and recorded every 4.8 ps
Kazlauskas, A. N. Serreqi, J. D. Schrag, E. Ziomek, M.
Cygler, Biochemistry 1994, 33, 3494–3500.
yielding 250 structures. Long-range electrostatic interactions
were calculated with particle mesh Ewald method and cut
off at 15 ꢂ, while the Lennard–Jones interactions were cut
off at 9.0 ꢂ. The root mean square deviations of the struc-
tures during the MD simulation were no more than 2 ꢂ,
(see Supporting Information, Figure S5. The four catalyti-
cally essential hydrogen bonds, His449 to SerOg and ester
oxygen, the two hydrogen bonds to the oxyanion, and the
catalytically non-productive hydrogen bond His449 to
[11] J. M. MancheÇo, M. A. Pernas, M. J. Martꢇnez, B.
Ochoa, M. L. Rffla, J. A. Hermoso, J. Mol. Biol. 2003,
332, 1059–1069.
[12] D. Ghosh, Z. Wawrzak, V. Z. Pletnev, N. Li, R. Kaiser,
W. Pangborn, H. Jçrnvall, M. Erman, W. L. Duax,
Structure 1995, 3, 279–288; V. Pletnev, A. Addlagatta,
Z. Wawrzak, W. Duax, Acta Crystallogr. D: Biol. Crys-
tallogr. 2003, 59, 50–56.
[13] K. Lundell, T. Raijola, L. T. Kanerva, Enzyme Microb.
ꢂ
Glu208 were deemed present if the N O distance was
<3.5 ꢂ and the N H O angle was >1208. A catalytically
Technol. 1998, 22, 86–93.
ꢂ ꢂ
[14] N. López, M. A. Pernas, L. M. Pastrana, A. Sꢆnchez, F.
Valero, M. L. Rffla, Biotechnol. Prog. 2004, 20, 65–73.
[15] S. H. Wu, Z. W. Guo, C. J. Sih, J. Am. Chem. Soc. 1990,
112, 1990–1995.
[16] I. J. Colton, S. N. Ahmed, R. J. Kazlauskas, J. Org.
Chem. 1995, 60, 212–217.
productive conformation was one that contained all four cat-
alytically essential hydrogen bonds and did not contain the
non-productive hydrogen bond. Table 2 lists the fraction of
catalytically productive conformation out of the 250 total
conformations.
[17] K. Nishizawa, Y. Ohgami, N. Matsuo, H. Kisida, H.
Hirohara, J. Chem. Soc. Perkin Trans. 2 1997, 1293–
1298.
[18] F. Texier-Boullet, A. Foucaud, Synthesis 1982, 165–166.
[19] M. S. Kharasch, R. A. Mosher, I. S. Bengelsdorf, J. Org.
Chem. 1960, 25, 1000–1006.
Acknowledgements
We thank the Natural Sciences and Engineering Research
Council of Canada for funding, the Minnesota Supercomput-
ing Institute for access to computer modeling hardware and
software, and Dr. Peter Bernhardt for initial molecular mod-
eling.
[20] Y.-F. Li, F. Hammerschmidt, Tetrahedron: Asymmetry
1993, 4, 109–120.
[21] A. Mezzetti, J. D. Schrag, C. S. Cheong, R. J. Kazlaus-
kas, Chem. Biol. 2005, 12, 427–437.
[22] S. Hur, T. C. Bruice, Proc. Natl. Acad. Sci. USA 2003,
100, 12015–12020.
[23] D. F. Sticke, L. G. Presta, K. A. Dill, G. D. Rose, J.
Mol. Biol. 1992, 226, 1143–1159.
[24] M. Botta, E. Cernia, F. Corelli, F. Manetti, S. Soro, Bio-
References
chim. Biophys. Acta 1997, 1337, 302–310.
[1] D. R. Feller, V. S. Kamanna, H. A. I. Newman, K. J.
Romstedt, D. T. Witiak, G. Bettoni, S. H. Bryant, D.
Conte-Camerino, F. Loiodice, V. Tortorella, J. Med.
Chem. 1987, 30, 1265–1267.
[2] G. Bettoni, F. Loiodice, V. Tortorelia, D. Conte-Camer-
ino, M. Mambri, E. Ferrannini, S. H. Bryant, J. Med.
Chem. 1987, 30, 1267–1270.
[3] S. N. Ahmed, R. J. Kazlauskas, A. H. Morinville, P.
Grochulski, J. D. Schrag, M. Cygler, Biocatalysis 1994,
9, 209–225.
[25] F. Manetti, D. Mileto, F. Corelli, S. Soro, C. Palocci, E.
Cernia, I. D’Acquarica, M. Lotti, L. Alberghina, M.
Botta, Biochim. Biophys. Acta 2000, 1543, 146–158.
[26] S. Mittal, S. Khanna, A. Roy, P. V. Bharatam, H. P. S.
Chawla, Enzyme Microb. Technol. 2005, 36, 232–238.
[27] K. Watanabe, T. Uno, T. Koshiba, T. Okamoto, Y.
Ebara, S. Ueji, Bull. Chem. Soc. Jpn. 2004, 77, 543–548.
[28] A. C. Sehgal, R. M. Kelly, J. Am. Chem. Soc. 2002, 124,
8190–8191.
[29] M. Holmquist, F. Haeffner, T. Norin, K. Hult, Protein
[4] S.-H. Wu, PhD thesis, University of Wisconsin, Madi-
Sci. 1996, 5, 83–88.
[30] P. Berglund, M. Holmquist, K. Hult, H. E. Hçgberg,
Biotechnol. Lett. 1995, 17, 55–60.
son, USA, 1987, p 120.
[5] S. K. Latypov, J. M. Seco, E. QuiÇoꢆ, R. Riguera, J.
Org. Chem. 1995, 60, 1538–1545; S. K. Latypov, J. M.
Seco, E. QuiÇoꢆ, R. Riguera, J. Am. Chem. Soc. 1998,
120, 877–882.
[31] P. Berglund, M. Holmquist, K. Hult, J. Mol. Catal. B:
Enzym. 1998, 5, 283–287.
[32] F. Secundo, G. Carrea, C. Tarabiono, P. Gatti-Lafranco-
ni, S. Brocca, M. Lotti, K.-E. Jaeger, M. Puls, T. Eggert,
J. Mol. Catal. B: Enzym. 2006, 39, 166–170.
[33] M. Holmquist, P. Berglund, Org. Lett. 1999, 1, 763–765.
[34] A. T. Brünger, P. D. Adams, G. M. Clore, W. L.
DeLano, P. Gros, R. W. Grosse-Kunstleve, J.-S. Jiang, J.
Kuszewski, M. Nilges, N. S. Pannu, R. J. Read, L. M.
Rice, T. Simonson, G. L. Warren, Acta Crystallogr. D:
Biol. Crystallogr. 1998, 54, 905–921.
[6] T. Miyazawa, S. Kurita, M. Shimaoka, S. Ueji, T.
Yamada, Chirality 1999, 11, 554–560.
[7] Review: P. D. de Maria, J. M. Sanchez-Montero, J. V.
Sinisterra, A. R. Alcantara, Biotechnol. Adv. 2006, 24,
178–194.
[8] J. J. Lalonde, C. Govardhan, N. Khalaf, A. G. Martinez,
K. Visuri, A. L. Margolin, J. Am. Chem. Soc. 1995, 117,
6845–6852.
[9] P. Grochulski, Y. Li, J. D. Schrag, F. Bouthillier, P.
Smith, D. Harrison, B. Rubin, M. Cygler, J. Biol. Chem.
1993, 268, 12843–12847.
[35] T. A. Jones, J.-Y. Zou, S. W. Cowan, Acta Crystallogr.
A: Crys. 1991, 47, 110–119.
[10] P. Grochulski, Y. Li, J. D. Schrag, M. Cygler, Protein
[36] P. Emsley, B. Lohkamp, W. G. Scott, K. Cowtan, Acta
Crystallogr. D: Biol. Crystallogr. 2010, 66, 486–501.
Sci. 1994, 3, 82–91; P. Grochulski, F. Bouthillier, R. J.
Adv. Synth. Catal. 2011, 353, 2529 – 2544
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2543

Ian J. Colton et al.
FULL PAPERS
[37] G. A. Kaminski, R. A. Friesner, J. Tirado-Rives, W. L.
Jorgensen, J. Phys. Chem. B 2001, 105, 6474–6487.
[38] J. R. Maple, M. J. Hwang, T. P. Stockfisch, U. Dinur, M.
Waldman, G. S. Ewig, A. T. Hagler, J. Comput. Chem.
1994, 15, 162–182.
[40] M. Rostkowski, M. H. M. Olsson, C. R. Søndergaard,
J. H. Jensen, BMC Struct. Biol. 2011, 11, 6.
[41] H. J. C. Berendsen, J. P. M. Postma, W. F. van Gunste-
ren, A. Dinola, J. R. Haak, J. Chem. Phys. 1984, 81,
3684–3690.
[39] W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W.
Impey, M. L. Klein, J. Chem. Phys. 1983, 79, 926–935.
2544
asc.wiley-vch.de
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2011, 353, 2529 – 2544