Different Active-Site Loop Orientation in Serine Hydrolases versus Acyltransferases

Article abstract of DOI:10.1002/cbic.201000693

Acyl transfer is a key reaction in biosynthesis, including synthesis of antibiotics and polyesters. Although researchers have long recognized the similar protein fold and catalytic machinery in acyltransferases and hydrolases, the molecular basis for the different reactivity has been a long-standing mystery. By comparison of X-ray structures, we identified a different oxyanion-loop orientation in the active site. In esterases/lipases a carbonyl oxygen points toward the active site, whereas in acyltransferases a NH of the main-chain amide points toward the active site. Amino acid sequence comparisons alone cannot identify such a difference in the main-chain orientation. To identify how this difference might change the reaction mechanism, we solved the X-ray crystal structure of Pseudomonas fluorescens esterase containing a sulfonate transition-state analogue bound to the active-site serine. This structure mimics the transition state for the attack of water on the acyl-enzyme and shows a bridging water molecule between the carbonyl oxygen mentioned above and the sulfonyl oxygen that mimics the attacking water. A possible mechanistic role for this bridging water molecule is to position and activate the attacking water molecule in hydrolases, but to deactivate the attacking water molecule in acyl transferases. All in the loop: A structure comparison of hydrolases and acyl transferases revealed a different conformation of the main chain in the oxyanion loop. Hydrolases point a C=O, whereas acytransferase point a NH, toward the active site. An X-ray structure of a hydrolase containing a sulfonate transition-state analogue show how this loop interacts with the attacking water.

Full text of DOI:10.1002/cbic.201000693

DOI: 10.1002/cbic.201000693
Different Active-Site Loop Orientation in Serine Hydrolases
versus Acyltransferases
Yun Jiang,[a] Krista L. Morley,[b] Joseph D. Schrag,*[c] and Romas J. Kazlauskas*[a]
Acyl transfer is a key reaction in biosynthesis, including synthe-
sis of antibiotics and polyesters. Although researchers have
long recognized the similar protein fold and catalytic machi-
nery in acyltransferases and hydrolases, the molecular basis for
the different reactivity has been a long-standing mystery. By
comparison of X-ray structures, we identified a different oxyan-
ion-loop orientation in the active site. In esterases/lipases a
carbonyl oxygen points toward the active site, whereas in acyl-
transferases a NH of the main-chain amide points toward the
active site. Amino acid sequence comparisons alone cannot
identify such a difference in the main-chain orientation. To
identify how this difference might change the reaction mecha-
nism, we solved the X-ray crystal structure of Pseudomonas
fluorescens esterase containing a sulfonate transition-state ana-
logue bound to the active-site serine. This structure mimics
the transition state for the attack of water on the acyl–enzyme
and shows a bridging water molecule between the carbonyl
oxygen mentioned above and the sulfonyl oxygen that mimics
the attacking water. A possible mechanistic role for this bridg-
ing water molecule is to position and activate the attacking
water molecule in hydrolases, but to deactivate the attacking
water molecule in acyl transferases.
Introduction
Acyltransferases (EC number 2.3) catalyze the transfer of an
acyl group from a donor, usually an ester or thioester, to an ac-
ceptor, usually an alcohol or amine, thereby forming esters or
amides (Scheme 1). Some acyltransferases are called thioester-
thesis use acyltransferases to catalyze macrocyclization or acy-
lation, Scheme 2. Macrocyclization reduces the conformational
freedom to ensure that the molecule is complementary to bio-
logical target and to reduce its susceptibility to proteases. Acy-
lation, usually with a hydrophobic acyl group, of the aglycone
or sugar moiety allow the antibiotics to interact with hydro-
phobic targets such as cell membranes. Acyltransferases also
catalyze key steps in amino acid biosynthesis. Homoserine O-
acetyltransferase (HTA) from Haemophilus influenzae catalyzes
the transfer of the acetyl group from acetyl-CoA to l-homoser-
ine to form O-acetylhomoserine Scheme 2.[2,3]
Acyltransferases also catalyze critical steps in the synthesis
of bacterial polyesters, triacylglycerols, and wax esters. Many
bacteria store excess carbon as insoluble granules of poly-b-hy-
droxybutyrate. The elongation step of this polyester synthesis
involves an acyl transfer of the growing polyester chain to an
added monomer.[4] In triacylglycerol or wax ester synthesis,
Scheme 1. Acyltransferases catalyze the transfer of an acyl group from a
donor (ester or thioester) to an acceptor (alcohol in this example). Both hy-
drolases and acyltransferase form an acyl–enzyme intermediate, but hydro-
lases transfer the acyl group to water, whereas acyltransferases transfer it to
an acceptor.
[a] Dr. Y. Jiang, Prof. Dr. R. J. Kazlauskas
Department of Biochemistry, Molecular Biology and Biophysics
Biotechnology Institute, University of Minnesota
1479 Gortner Avenue, Saint Paul, MN, 55108 (USA)
Fax: (+1)612-625-5780
ases for their homology to fatty acyl thioesterases. In fatty acid
biosynthesis, these thioesterases catalyze hydrolysis, but the
thioesterase domains in polyketide biosynthesis and nonribo-
somal peptide biosynthesis can catalyze either hydrolysis or
acyl transfer. The acyl transfer reaction is similar to the hydroly-
sis reaction catalyzed by carboxyesterases and lipases (EC
number 3.1.1). The difference is that acyl transfer involves
transfer to an acceptor, whereas in hydrolysis the acceptor is
water. Acyl transfer is kinetically controlled; the thermodynami-
cally favored product is hydrolysis.
E-mail: rjk@umn.edu
[b] Dr. K. L. Morley
Department of Chemistry, McGill University
Montrꢀal, Quꢀbec, H3A 2K6 (Canada)
[c] Dr. J. D. Schrag
National Research Council Biotechnology Research Institute
6100 Royalmount Avenue, Montrꢀal, Quꢀbec, H4P 2R2 (Canada)
Fax: (+1)514-496-5143
Acyltransferases catalyze critical steps in antibiotic biosyn-
thesis and other pathways such as amino acid biosynthesis.[1]
Both polyketide biosynthesis and non-ribosomal peptide syn-
E-mail: joe@bri.nrc.ca
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201000693.
768
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2011, 12, 768 – 776

Loop Orientation in Hydrolases and Acyltransferases
catalyzed by Srf TE domain in vitro.[12] Homoserine
acetyltransferase and DEBS TE catalyze hydrolysis of
acetyl CoA and N-acetylcysteamine thioester deriva-
tives in the absence of an acceptor.[3,13] Some in vitro
experiments use thioesterases excised from large
multidomain complexes and show more hydrolysis
than expected. Addition of solvents and surfactants
also influences the relative amount of hydrolysis.
This overlapping catalytic activity in esterases/lipas-
es and acyltransferases is due to their similar 3D
structures and similar catalytic machinery. Both adopt
the a/b-hydrolase fold.[14] A few esterases and acyl-
transferases adopt other folds such as the b-lacta-
mase fold adopted by esterase B from Burkholderia
gladioli[15] and simvastatin acyltransferase,[16] but
these other folds are not considered in this paper.
Both esterases/lipases and acyltransferases in the a/
b-hydrolase-fold class contain a Ser-His-Asp(Glu) cata-
lytic triad, an oxyanion hole and follow a ping-pong
bi–bi reaction mechanism involving an acyl–serine
enzyme intermediate. In both cases, catalysis starts
by the serine nucleophile attacking the substrate’s
Scheme 2. Acyltransferases catalyze key steps in antibiotic and amino acid synthesis.
A) Surfactin synthase thioesterase domain (SrfTE) catalyzes intramolecular cyclization of
peptidyl carrier protein thioester to form macrolactone cyclic heptapeptide. B) As part of
methionine biosynthesis homoserine O-acetyltransferase (HTA) catalyzes the transfer of
the acetyl group from acetyl-CoA to l-homoserine to form O-acetylhomoserine.
acytransferases add the acyl group from acyl–coenzyme A to a
diacylglycerol to make a triacylglycerol, or to a long-chain alco-
hol to make a wax ester.[5] Cholesterol transport and lipopro-
tein assembly also involves acyltransferases. Inhibitors of N-
myristoyltransferase might alleviate sleeping sickness.[6]
carbonyl carbon and the formation of an acyl–enzyme inter-
mediate. The next step differs in the two enzyme classes: in es-
terases/lipases: a water nucleophile attacks acyl–enzyme inter-
mediate, whereas in acyltransferases an alcohol nucleophile (or
other acyl acceptor) attacks acyl–enzyme intermediate.
Acyltransferases are also used in biocatalysis for the chemo-
enzymatic synthesis of natural product analogues. For exam-
ple, Walsh and co-workers used a cloned thioesterase to cata-
lyze the macrocylization of peptides to create small libraries of
macrolactams.[7] Such reactions would have yielded mixtures of
different cyclization products without an enzyme. Later the
same group used a different thioesterase to regioselectively
acylate vancomycin or teicoplanin variants.[8] In another acyla-
tion example, Tang’s group used the acyltransferase LovD in
the gram-scale chemoenzymatic synthesis of simvastatin, a
cholesterol-lowering drug that is a semisynthetic analogue of
the natural product lovastatin. This procedure allowed the use
of water as the solvent and avoided multiple protection and
deprotection steps.[9] Green and co-workers used choline acyl-
transferases to make potential inhibitors of choline acyltrans-
ferases themselves.[10]
Despite the X-ray crystal structures of eight acyltransferases,
including structures with bound substrate analogues, the
molecular basis of acyltransferase activity versus hydrolysis
remains elusive. The key difference between these reactions is
choosing between water and alcohol (or amine) as the nucleo-
phile. There are two ways to favor acyltransfer over hydrolysis:
first, decrease the ability of water to act as a nucleophile or
second, increase the ability of alcohol to act as a nucleophile.
Both approaches can work simultaneously. In two cases crystal-
lographers suggested that the active site of acyltransferases is
hydrophobic and excludes water (Srf TE, Mycobacterium anti-
gens), but in three cases (HiHAT, LiHAT, fengycin TE) crystallog-
raphers did not mention any evidence for a water-excluding
mechanism. Most structures show evidence for specific interac-
tions between the alcohol and acyltransferase that favor acyl-
transfer over hydrolysis. However, the alcohol differs for each
acyltransferase, so the structural features that favor its binding
will also differ. We did not study the alcohol binding contribu-
tion to acyltransfer in this paper.
Esterases/lipases and acyltransferases catalyze both hydroly-
sis and acyl transfer, but esterases/lipases favor hydrolysis
(transfer of the acyl group to water), whereas acyltransferases
favor transfer to an acceptor other than water. In nonaqueous
solvents, esterases and lipases favor acyl transfer to an accept-
or over hydrolysis. Lyophilized powders of esterases or lipases
suspended in nonaqueous solvents, typically ionic liquids or
nonpolar organic solvents, catalyze acyl transfer.[11] Examples
include the resolution of alcohols by enantioselective acylation,
regioselective acylation of sugars, and the synthesis of poly-
mers by ring-opening polymerization of lactones.
Our hypothesis is that, besides binding the nucleophile,
which is specific for each nucleophile, there is also a common
mechanism for deactivation of water as a nucleophile in acyl-
transferases. Our approach is to compare the X-ray crystal
structures of related hydrolases and acyltransferases to identify
the structural differences, especially in the active-site region
that binds the nucleophilic water or alcohol. We found a differ-
ence in the main-chain orientation of the oxyanion loop in the
active site and propose a mechanism for how this different ori-
entation can deactivate water as a nucleophile.
Hydrolysis is a common side reaction of acyltransferases, es-
pecially in the absence of an acceptor. For example, ~30% hy-
drolysis accompanied the macrocyclization of a natural peptide
ChemBioChem 2011, 12, 768 – 776
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
769

R. J. Kazlauskas et al.
The central residues in a b-turn can adopt different confor-
mations. The type-I b-turn contains Fi+1 =À608, yi+1 =À308,
Fi+2 =À908, yi+2 =08, whereas the type-II b-turn contains
Fi+1 =À608, yi+1 =1208, Fi+2 =808, yi+2 =08. Conformations
can vary within Æ308 of these ideal values. In PFE these angles
are À668, 1358, 878, À218 and thus correspond to a type-II
turn (i+1=Trp28, i+2=Leu29). In HiHAT, these angles are
À488, À498, À1098, 148 and correspond to a type-I turn (i+1=
Leu49, i+2=Thr50. The consequence of this difference is that
the carbonyl groups of PFE Trp28 and HAT Leu49 point in
opposite directions: PFE Trp28 carbonyl points into the active
site, whereas HAT Leu49 carbonyl points away from the active
site.
Results
Structure differences between Haemophilus influenzae
homoserine acyltransferase (HiHAT) and Pseudomonas
fluorescens esterase (PFE)
Although the amino acid sequences of HiHAT and PFE differ
signficantly (only 13% sequence identity), their 3D structures
(PDB IDs: 2b61 and 1va4, respectively) are very similar. The Z-
score for the structural overlay by using DALI[17] is 23.8 and
root-mean-square deviation between the atoms in the two
structures is 3.1 ꢁ over 257 amino acids.
The major difference in the active sites is the oxyanion loop
conformation. This loop occurs after strand b3 in the a/b-hy-
drolase fold and contains one residue that donates a main-
chain NH to stabilize the oxyanion intermediate during cataly-
sis. This residue, called the first oxyanion residue in this paper,
is Leu49 in HiHAT and Trp28 in PFE. In HiHAT this loop adopts
a type-I b-turn, whereas in PFE it adopts a type-II b-turn.
b-Turns reverse the direction of the peptide chain and con-
sist of four amino acid residues; the first and last residues (i
and i+3) form a hydrogen bond between C=O (i) to NH
(i+3),[18] Figure 1. This hydrogen bond is similar to the one be-
Different oxyanion loop conformation in GX-class hydrolases
versus acyltransferases
We hypothesize that this difference in oxyanion turn orienta-
tion for PFE and HiHAT is a common feature that distinguishes
esterases from acyltransferases. To test this hypothesis we
made a more extensive comparison of structures within closely
related esterases/lipases and acyltransferases. For the hydrolas-
es, we focused on esterases/lipases within the a/b-hydrolase
superfamily and further narrowed our comparison to those
with a GX-class oxyanion loop. The lipase engineering data-
base subdivides lipases according the oxyanion loop orienta-
tion into GX, GGGX, and Y classes,[19] in which G represents gly-
cine, X represents any amino acid, and Y represents tyrosine.
The oxyanion loop orientation in acyltransferases (see below)
is similar to GX family hydrolases, but the other two classes
have significantly different orientations of the oxyanion loops.
The GGGX class has a much larger space, whereas the Y class
uses the tyrosine side chain to stabilize the oxyanion so that
the main-chain orientation differs significantly. As mentioned
above, we excluded any structures such as closed conforma-
tions of lipases in which there was doubt as to whether it was
a catalytically active conformation. We included all hydrolases
in the structural classification of proteins (SCOP) database[20]
that fit these criteria. We found thirty-two X-ray structures of
hydrolases that fit these criteria; these are listed in Table S1 in
the Supporting Information.
Figure 1. b-turns. A) Idealized type-I b-turn (cyan carbons) overlaid on an
idealized type-II b-turn (green carbons). The amino acids are XxxAlaAlaXxx,
where Xxx indicates an incompletely specified residue since only part of the
i and i+3 residues are shown. The labels are near the Ca of each residue;
dotted lines indicate hydrogen bonds between the carbonyl of residue i and
the NH of residue i+3. B) Superimposed X-ray crystal structures of the oxy-
anion loop regions of PFE (green carbons, a type-II b-turn) and HiHAT (cyan
carbons, a type-I b-turn). The structures were superimposed by pair fitting
of the active site serine and histidine (not shown in Figure for clarity). In this
view, the active site lies behind the plane of the paper so the carbonyl of
i+1 in PFE (green) points toward the active site, whereas for HiHAT (cyan
carbons) the carbonyl of i+1 points away from the active site. In both cases
the NH of i+1 (circled with a dashed line) points toward the active site be-
cause it forms a hydrogen bond with the oxyanion intermediate.
The oxyanion loop in most of these hydrolases adopts a
type-II b-turn and all but one point the carbonyl group of the
i+1 residue toward the active site, Table S1. Among thirty-two
GX-class hydrolases, twenty-one have angles are within 308 of
the ideal values defined above for type-II b-turns, three have
angles deviate slightly more than 308 from the ideal values,
seven are not type-II b-turn because the last two angles (Fi+2
,
yi+2) deviate significantly from those for a type-II b-turn. In all
thirty-one of these structures, the first two dihedral angles in
the definition of a type-II b-turn (Fi+1, yi+1) are similar. The
Fi+1 angles range between À828 and À448 and the yi+1
angles range between 1128 and 1558. Because of this similarity
in the dihedral angles for i+1, the carbonyl group of this resi-
due points toward the active site in all thirty-one structures.
The backbone NH of this residue, whose catalytic role is stabili-
zation of the oxyanion, also points toward the active site. The
tween adjacent antiparallel b strands; hence the name b-turn.
The two central residues in a b-turn (second and third residues
called i+1 and i+2, respectively) lack hydrogen bonds be-
tween the main-chain carbonyl and NH groups. This lack of hy-
drogen bond is critical for catalysis in ester hydrolysis and acyl
transfer because the i+1 NH group stabilizes the oxyanion in-
termediate by donating a hydrogen bond.
770
www.chembiochem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2011, 12, 768 – 776

Loop Orientation in Hydrolases and Acyltransferases
one exception to this generalization is human fatty acid syn-
thetase thioesterase domain, which shows a type-I b-turn for
the oxyanion turn[21] This exception is rationalized in the Sup-
porting Information.
Acyltransferases fall into several superfamilies within the
SCOP database, but only acyltransferases that have the a/b-
hydrolase fold are included in this comparison. For example,
homoserine O-succinyl transferase from Bacillus cereus (PDB ID:
2ghr) is not included because its structure belongs to the
class-I glutamine amidotransferase-like superfamily. Similarly,
histone acetylases are not included because most structures
belong to the acyl-CoA N-acyltransferases superfamily. Three
acyltransferases within the a/b-hydrolase fold superfamily are
not included in Table S2 because their oxyanion loops are dis-
torted. Two thioesterases of polyketide biosynthesis (erythro-
mycin thioesterase and pikromycin thioesterase) contain a
two-residue insertion in the oxyanion loop, which significantly
changes its orientation, see the Supporting Information for
details. Myristoyl-ACP-specific thioesterase is also excluded be-
cause the X-ray structure[22] did not definitively identify the
oxyanion loop. The most likely oxyanion loop is too far from
the active site: 8.6 ꢁ from Og of Ser114 to the backbone nitro-
gen of Phe44. (The corresponding distance in PFE is 4.8 ꢁ.) Cat-
alysis might require a conformational change or a bridging
water molecule. Table S2 lists the eight acyltransferases in the
a/b-hydrolase fold superfamily that fit our criteria. These in-
clude three mycolyl transferases (Mycobacterium tuberculosis
antigens), two thioesterases from polyketide antibiotic biosyn-
thesis that catalyze cyclization of surfactin or fengycin, two ho-
moserine acyltransferases and an acetyl transferase from the b-
lactam antibiotic biosynthesis.
Figure 2. Dihedral angles of the oxyanion residue (i+1 in the turn) in hydro-
lases and acyltransferases. The hydrolases (diamonds) have Fi+1 =À62Æ88,
yi+1 =134Æ98, which points the carbonyl oxygen of these residues toward
the active site. The acyltransferases (crosses) have Fi+1 =À54Æ88, yi+1
=
À40Æ108, which points the carbonyl oxygen of these residues away from
the active site. The average difference in yi+1 is 1748, which corresponds to
an opposite orientation for the carbonyl oxygen of these residues. One hy-
drolase, labeled 1xkt, lies among the acyltransferases and is not included in
the averages. The Supporting Information suggests an explanation for this
exception. The dotted lines indicate angles within 308 of the ideal values for
b-turns. (This Figure shows that the dihedral angles of the oxyanion residue
are in the range for a type-II b-turn for 30 of the 31 hydrolases. Only 23 of
the 31 hydrolases also have the dihedral angle for the next amino acid resi-
due in the range for a type-II b-turn. The orientation of the next amino acid
does not affect orientation of the key carbonyl oxygen.)
The oxyanion turns in all eight acyltransferases in Table S2
are either type-I b-turns or very close to a type-I b-turn. In par-
ticular, the dihedral angles of the oxyanion residue are similar
in each case: the Fi+1 angles range from À708 to À468 and
the yi+1 angles range from À528 to À248. The conformation of
the oxyanion residues maintains the catalytically essential NH
in position to hydrogen bond to the oxyanion intermediate.
However, this conformation points the carbonyl oxygen of this
residue away from the active site. This different orientation of
the carbonyl group is the major structural difference between
hydrolases and acyltransferases.
X-ray crystal structure of a sulfonate transition-state
analogue bound to PFE to mimic hydrolysis
The key difference between hydrolysis and acyl transfer is the
choice between water or alcohol as the nucleophile. A phos-
phonate transition-state analogue mimics the attack of alcohol
on the acyl–enzyme, whereas a sulfonate transition-state ana-
logue mimics the attack of water on the acyl–enzyme. Phos-
phonates contain an -OR moiety that mimics the alcohol of
the ester, whereas sulfonates contain a sulfonyl oxygen to
mimic the attacking water, Scheme 3. Crystallographers have
solved many X-ray crystal structures of lipases and esterases
containing bound phosphonate transition-state analogues, but
not of a sulfonate transition-state analogue bound to a GX-
class hydrolase. For this reason, we synthesized butane-2-sul-
fonic acid 4-nitrophenyl ester as an irreversible inhibitor of PFE
and solved its X-ray crystal structure bound to PFE. (Two previ-
ous X-ray structures of a sulfonate transition-state analogues
bound to an esterase[23] are catalytically not productive orienta-
tions because some of the key hydrogen bonds are missing, or
the transition-state analogue orients in with the acyl mimic in
A graphic comparison of conformations of the oxyanion resi-
due (i+1 of the b-turn) shows a dramatic difference between
the hydrolases and the acyltransferases, Figure 2. The Fi+1
angles are similar for both: an average of À62Æ88 for the hy-
drolases in Table S1 (excluding hydrolase 1xkt) and and aver-
age of À54Æ88 for the acyltransferases in Table S2. This simi-
larity places the backbone amide NH and the Ca of i+1 in sim-
ilar orientations and allows this amide NH to contribute to cat-
alysis in both classes of enzymes. In contrast, the yi+1 angles
differ. For hydrolases, this angle averages 134Æ98, whereas for
acyltransferases it averages À40Æ108. This 1748 difference cor-
responds to an opposite orientation for the carbonyl oxygen
of this residue.
ChemBioChem 2011, 12, 768 – 776
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
771

R. J. Kazlauskas et al.
In the case of attack of water
on an acyl–enzyme intermediate,
this bridging water interaction is
expected to be stronger than it
is for the sulfonyl transition-state
analogue. The bridging water
molecule will not interact with a
sulfonyl oxygen, but rather with
an attacking water molecule.
The oxygen of this water mole-
cule will be more negatively
charged than a sulfonyl oxygen
because the active-site histidine
acts as base to deprotonate it.
An X-ray crystal structure of a
sulfonate transition-state ana-
logue bound to an acyltransfer-
ase (fengycin thioesterase) un-
Scheme 3. Mechanism for hydrolysis of an ester, R1C(O)OR2, by a serine esterase. Attack of the serine on the ester
forms the first tetrahedral intermediate, Td1. A phosphonate linked to the serine mimics this first intermediate.
The R1 on the phosphonate corresponds to the R1 on the acyl part of the ester and R2 on the phosphonate corre-
sponds to the R2 on the alcohol part of the ester. Loss of the alcohol, HOR2, from Td1 gives the acyl–enzyme inter-
mediate. Attack of water on the acyl–enzyme forms the second tetrahedral intermediate, Td2. A sulfonate linked
to the serine mimics this second tetrahedral intermediate. The R1 on the sulfonate corresponds to R1 on the acyl
part of the ester. One of the sulfonate oxygens mimics the oxyanion oxygen of Td2, and the other sulfonate oxy-
gen mimics the attacking water of Td2. Loss of the serine from Td2 gives the product acid. For simplicity, the draw-
ing omits protonation and deprotonation steps.
the alcohol region.) Details of the crystallization, data collec-
tion, and refinement are in the Supporting Information.
Alignment of the complex and wild-type structures reveals
only minor changes to the active site. The 2-butyl substituent
points toward the solvent, Figure 3. Detailed analysis of this
orientation and implications for enantioselectivity will be re-
ported elsewhere. The inhibitor mimics a catalytically produc-
tive orientation because it contains all of the five hydrogen
bonds necessary for catalysis. The sulfonyl oxygen in the oxy-
anion hole is hydrogen bonded to the NH of both Met95 and
Trp28 (2.9 and 3.0 ꢁ, respectively), whereas the second sulfonyl
oxygen is weakly hydrogen bonded to His251 Ne2 (3.3 ꢁ) and
to the bridging water molecule (3.2 ꢁ). The unit cell contains
six protein molecules; chains A and D show this bridging
water molecule, while the other chains do not, likely due to
disorder. This bridging water molecule also hydrogen bonds to
the main-chain carbonyl oxygen of Trp28 (2.5 ꢁ). The catalytic
Ser94 Og is hydrogen bonded to the His251 Ne2 (3.2 ꢁ) and
Asp222 Od2 is hydrogen bonded to His251 Nd1 (2.8 ꢁ; also
not shown in Figure 3 for clarity). Figure 3B shows a schematic
of the hydrogen bonds between the sulfonate transitions-state
analogue and the protein, whereas panel C shows how these
hydrogen bonds would look in the tetrahedral intermediate
during hydrolysis. The X-ray crystal structure does not reveal
the positions of the hydrogen atoms; all hydrogen bonds are
inferred from the close distances of the oxygen or nitrogen
atoms.
Figure 3. X-ray crystal structure of a sulfonate transition-state analogue
bound to the catalytic serine of esterase from Pseudomonas fluorescens
suggests a mechanistic role for a main-chain carbonyl oxygen of the oxyan-
ion loop. A) The active site of PFE showing key hydrogen bonds to the sulfo-
nate transition-state analogue covalently linked to Ser94. One sulfonyl oxy-
gen, partly obscured by the R group of the sulfonate in this view, mimics
the oxyanion oxygen and accepts hydrogen bonds from the main-chain NH
of Met95 and Trp28 (2.9 and 3.0 ꢁ, respectively). Another sulfonyl oxygen
mimics the attacking water molecule and accepts hydrogen bonds from the
catalytic His251 and a bridging water molecule (ball; 3.3 and 3.2 ꢁ, respec-
tively). This bridging water molecule also hydrogen bonds to the main-chain
carbonyl oxygen of tryptophan 28 (2.5 ꢁ). B) A schematic diagram of the X-
ray crystal structure in panel a showing the hydrogen bonding pattern.
C) Schematic diagram of the tetrahedral intermediate that corresponds to
the sulfonate mimic in panel A and B. The attacking water molecule (bond-
ed to the reacting carbon) makes hydrogen bonds to both the active site
His251 and the bridging water molecule.
To confirm that the water molecule bridging the sulfonyl
transition-state analogue and the oxyanion loop is stable at
that location, we used a molecular dynamics simulation (Fig-
ure 4). By starting from the X-ray crystal structure (subunit D)
placed in a box of water molecules, molecular dynamics simu-
lated movement of the complex for 1.2 ns. The bridging water
molecule remained in position and maintained the two hydro-
gen bonds (average OÀO distance ~2.9 ꢁ).
772
www.chembiochem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2011, 12, 768 – 776

Loop Orientation in Hydrolases and Acyltransferases
ecule can act as an additional base. Although in prin-
ciple, a water molecule can be either a hydrogen-
bond donor or acceptor, the interaction of this bridg-
ing water molecule with the carbonyl oxygen of
Trp28 sets its role as a base. The carbonyl oxygen of
Trp28 can only accept a hydrogen bond, and thus,
the bridging water donates a hydrogen bond. Be-
cause its partial negative charge has increased, the
bridging water molecule will now accept a hydrogen
bond from the attacking water. Another role for the
bridging water molecule during hydrolysis might be
to aid positioning the attacking water molecule in a
catalytically productive orientation.
In contrast to acyl transferases, the bridging water
molecule in acyltransferases might hinder the hydrol-
ysis step. First, the carbonyl in hydrolases and the NH
in acyltranferases are in slightly different locations,
which places the bridging water molecule further
from the serine in acyltransferases than in hydrolases.
The bridging water molecule might be too far away
to interact with the attacking water molecule or it
might position the attacking water molecule too far
from the carbonyl carbon to attack effectively.
Second, the NH donates a hydrogen bond to the
bridging water so that it, in turn, can donate a hydro-
gen bond to the attacking water, Figure 5. This acidic
interaction decreases the nucleophilicity of the at-
tacking water and might slow down hydrolysis.
Figure 4. Molecular dynamics simulation of a bound water molecule in the active site of
PFE. The X-ray crystal structure of some subunits showed a water bridging from the tran-
sition-state analogue to the oxyanion loop with hydrogen bonds. The molecular dynam-
ics simulation follows these hydrogen bonds. The distances from the bound water to sul-
fonyl transition-state analogue oxygen (shown in black, average value is 2.95 ꢁ) and
Trp28 carbonyl oxygen (shown in gray, average value is 2.87 ꢁ) remain near 2.9 ꢁ, which
indicates stable hydrogen bonds.
Finally, during acyl transfer in acyl transferases, the
fortunately shows a catalytically nonproductive orientation and
thus, provides little insight into the mechanism.[24] Fengycin
thioesterase catalyzes the macrocyclization of a peptide, but
the transition-state analogue was phenylmethylsulfonyl fluo-
ride. The phenylmethylsulfonyl moiety bound to the active-site
serine, but imperfectly mimicked the transition state, likely
because the phenylmethyl moiety imperfectly mimicked the
substrate peptide. One sulfonyl oxygen formed the expected
two hydrogen bonds in the oxyanion hole, but the active-site
histidine adopted a catalytically non-productive orientation. A
bridging water molecule accepted a hydrogen bond from the
NH of the oxyanion loop (Ser31; NÀO distance 3.3 ꢁ). The
bridging water molecule did not donate a hydrogen bond the
sulfonate oxygen, which mimics the attacking water (OÀO dis-
tance 4.2 ꢁ). One can conclude that a bridging water molecule
likely exists in the active site during catalysis by this acyltrans-
ferase, but one cannot say anything about its role in catalysis.
bridging water molecule has no effect. The acceptor alcohol or
amine displaces the bridging water molecule from the active
site, so it does not aid or hinder catalysis, Figure 5. The accept-
ors are larger than a water molecule and therefore displace
the bridging water molecule. X-ray crystal structures of phos-
phonates bound to the active site of both hydrolases (e.g., Burk-
holderia cepacia lipase,[25] Pseudomonas aeruginosa lipase,[26]
Bacillus subtilis lipase A,[27] dog gastric lipase[28] and Rhizomucor
miehei lipase[29]) and acyltransferases (e.g., Mycobacterium anti-
gen[30]) show either no water molecule or a water molecule
that is too far to form a hydrogen bond to the alcohol oxygen.
Discussion
Other crystal structures of hydrolases have also identified sec-
ondary base interactions to the attacking water molecule in-
cluding several that involve a bridging water molecule. In the
glutamic peptidase family, the hydrolytic water hydrogen
bonds to the catalytic glutamate base and to the carbonyl
oxygen Oe1 of a nearby glutamine.[31] In a glycoside hydrolase,
the hydrolytic water hydrogen bonds with both the catalytic
aspartate and an adjacent tyrosine phenolic oxygen that helps
position the water molecule.[32] Three examples involve a
bridging water molecule. Potato epoxide hydrolase and haloal-
kane dehalogenase, which both have the same a/b-hydrolase
fold as esterases and acyltransferases, show a bridging interac-
tion that is the same as the one proposed for esterases in the
current paper. In epoxide hydrolase, a bridging water molecule
Mechanistic significance of the type-I versus type-II
orientation of the oxyanion turn
The interaction with the bridging water molecule might con-
tribute to catalysis, Figure 5. In a hydrolase, the bridging water
molecule could act as a base by accepting a hydrogen bond
from the attacking water molecule. This additional basic inter-
action could increase the reactivity of the attacking water mol-
ecule. The role of the active-site histidine as a base is well es-
tablished; the current proposal is that the bridging water mol-
ChemBioChem 2011, 12, 768 – 776
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
773

R. J. Kazlauskas et al.
proteases are a second example of hydrolases that show no
secondary base interaction. The X-ray crystal structures pro-
teases of acyl–enzyme intermediates in several serine proteas-
es have identified the likely hydrolytic water.[39] This water
hydrogen bonds to Ne2 of the catalytic histidine, but not to
anything else.
Other hydrolases can use alternative mechanisms to activate
water. Lipase from Candida antarctica has a water tunnel that
connects the solvent to the active site. Blocking this tunnel by
site-directed mutagenesis decreased the relative amount of
hydrolysis versus transesterification by approximately sixfold.[40]
This tunnel mechanism could be more important for lipases,
which might be partly buried in a lipid during catalysis. Subtili-
sins also contain a water tunnel.[41]
If a secondary base interaction can activate a water for reac-
tion, then a secondary acid interaction should deactivate a
water for reaction. In acyltransferases the oxyanion turn orients
the central amide bond in an opposite manner so an NH, not
a carbonyl oxygen point toward the active site. Although no
structural information is available on the attacking water mole-
cule in these enzymes, we hypothesize that the situation is
similar to GX-class hydrolases. However, the NH donates a hy-
drogen bond to the bridging water, which in turn donates a
hydrogen bond to the attacking water. This donation of a hy-
drogen decreases the nucleophilicity of the attacking water.
This interaction could be partly responsible for the different
chemical reactivity of the hydrolases versus acyltransferases.
Alternative hypotheses for the significance of the different
oxyanion loop orientation to the hydrolase versus acyltransfer-
ease mechanism are possible. For example, the different oxy-
anion loop orientation points the dipole of the carbonyl in
different directions in hydrolases and acyltransferases. The
strength of our work is to identify the structural difference
between the two classes of enzymes. Our suggestions for the
role of the bridging water molecule are consistent with the
known structure and mechanisms, but it does not rule out al-
ternative explanations.
Figure 5. Hypothesis for how the different conformation of the oxyanion
loop in esterases and acyltransferases can activate or deactivate the attack-
ing water via a second water molecule. In esterases, the main-chain carbonyl
oxygen acts as a base via a bridging water molecule to activate the attack-
ing water molecule. In acyltransferases, the NH acts as an acid via a bridging
water molecule to deactivate the attacking water molecule. For the acyl
transfer reaction, the attacking nucleophile is an alcohol, which is larger
than water. This alcohol displaces the bridging water molecule thereby elim-
inating any activating or deactivating effects.
hydrogen bonds to the main-chain carbonyl oxygen of Phe33
and the attacking water,[33] whereas in the haloalkane dehalo-
genase this bridge is between main-chain carbonyl oxygen of
Asn38 and the attacking water.[34] In the second case, molecu-
lar dynamics simulations also suggest that this bridging water
molecule has an important role in catalysis. In a penicllinase,
the attacking water hydrogen bonds to both the base (Og of
Ser87) and a bridging water molecule linked to Og of
Ser110.[35] Recent modeling and mutagenesis indicated that a
water network contributes to the amidase activity of a Bacillus
esterase.[36] Using a bridging water molecule in addition to the
hydrolytic water molecule is disfavored by entropy, but the en-
thalpic advantage of the hydrogen bonds formed likely com-
pensate for the entropic cost.[37]
Previous research had identified that the oxyanion loop can
contribute the preference for acyl transfer, but had not identi-
fied how. Site-directed mutagenesis in the oxyanion loop of
surfactin thioeserase increased the relative amount of hydroly-
sis over acyl transfer.[12] Wild-type enzyme showed 71% acyl
transfer, but the Pro29Gly variant showed only 23% acyl trans-
fer, the rest was hydrolysis. This amino acid substitution likely
changed the conformation of the oxyanion turn due to the dif-
ferent conformational preferences of proline versus glycine,
but the nature of the conformational change was not estab-
lished. In cyclization reactions, the higher effective molarity of
the nucleophile due to the folded conformation of peptide or
polyketide can favor acyl transfer over hydrolysis, but many
acyl transfers do not involve cyclization.
Stehle and co-workers[42] made homology models of plant
acyltransferases[43] but, after careful analysis, concluded that
“there is no reliable explanation” for the reactivity differences
between hydrolases and acyltransferases. The homology
models were based on hydrolases and thus contained a hydro-
lase-like oxyanion loop orientation. Our proposed molecular
Secondary base interaction is not essential for hydrolases ac-
tivity because some hydrolases show no evidence of a secon-
dary base interaction. One previous X-ray crystal structure of a
lipase did contain a sulfonate transition-state analogue,[38] but
this was not a GX-class hydrolase. Candida rugosa lipase be-
longs to the GGGX class of lipases, the oxyanion loop of which
is farther from the active-site serine and tetrahedral intermedi-
ate to interact with them directly or by way of a bridging
water molecule. There was no evidence of a bridging water
molecule in this structure, so the proposed secondary base
interaction might not be essential for catalysis. Several serine
774
www.chembiochem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2011, 12, 768 – 776

Loop Orientation in Hydrolases and Acyltransferases
basis involves main-chain interactions, which are difficult to
find by using amino acid sequence comparisons. There is no
clear link between amino acid sequence and the types-I or II
conformation in a b-turn. For this reason, a bioinformatics
approach would not identify sequence motifs characteristic of
hydrolases or acyltransferases. We previously identified the im-
portance of main-chain interactions for perhydrolysis activity.[44]
Moving the same oxyanion loop closer to the active site could
create a perfect fit for hydrogen peroxide and thus, enhance
perhydrolysis.
Keywords: hydrolases · oxyanion loops · structure–activity
relationships · transferases · X-ray diffraction
[1] J. Grꢃnewald, M. A. Marahiel, Microbiol. Mol. Biol. Rev. 2006, 70, 121–
146; F. Kopp, M. A. Marahiel, Nat. Prod. Rep. 2007, 24, 735–749.
[2] I. A. Mirza, I. Nazi, M. Korczynska, G. D. Wright, A. M. Berghuis, Biochem-
istry 2005, 44, 15768–15773.
[3] T. L. Born, M. Franklin, J. S. Blanchard, Biochemistry 2000, 39, 8556–
8564.
[4] J. Stubbe, J. Tian, A. He, A. J. Sinskey, A. G. Lawrence, P. Liu, Ann. Rev.
Biochem. 2005, 74, 433–480.
[5] M. Wꢄltermann, T. Stçveken, A. Steinbꢃchel, Biochimie 2007, 89, 230–
242.
Experimental Section
[6] J. A. Frearson, S. Brand, S. P. McElroy, L. A. T. Cleghorn, O. Smid, L. Stoja-
novski, H. P. Price, M. L. S. Guther, J. S. Torrie, D. A. Robinson, I. Hallybur-
ton, C. P. Mpamhanga, J. A. Brannigan, A. J. Wilkinson, M. Hodgkinson,
R. Hui, W. Qiu, O. G. Raimi, D. M. F. van Aalten, R. Brenk, I. H. Gilbert,
K. D. Read, A. H. Fairlamb, M. A. J. Ferguson, D. F. Smith, P. G. Wyatt,
Nature 2010, 464, 728–732.
[7] J. W. Trauger, R. M. Kohli, H. D. Mootz, M. A. Marahiel, C. T. Walsh, Nature
2000, 407, 215–218; R. M. Kohli, M. D. Burke, J. Tao, C. T. Walsh, J. Am.
Chem. Soc. 2003, 125, 7160–7161.
[8] R. G. Kruger, W. Lu, M. Oberthꢃr, J. Tao, D. Kahne, C. T. Walsh, Chem. Biol.
2005, 12, 131–140.
[9] X. Xie, K. Watanabe, W. A. Wojcicki, C. C. C. Wang, Y. Tang, Chem. Biol.
2006, 13, 1161–1169; X. Xie, Y. Tang, Appl. Environ. Microbiol. 2007, 73,
2054–2060.
(Æ)-Butane-2-sulfonic acid 4-nitrophenyl ester: (S)-sec-butyl etha-
nethioate[45] was oxidized to the sulfonic acid and isolated as the
sodium salt. Chlorination with thionyl chloride and addition of p-ni-
trophenol and Et3N yielded the desired p-nitrophenyl sulfonate
ester. The Supporting Information contains synthetic details.
X-ray crystal structure of (Æ)-butane-2-sulfonyl bound to PFE:
Incubation of wild-type PFE for 24 h with a tenfold excess of (Æ)-
butane-2-sulfonic acid 4-nitrophenyl ester inhibited 95% of the
hydrolytic activity. The PFE–inhibitor complex and mutants formed
single crystals under conditions similar to those for wild-type
PFE:[46] 1.7m (NH4)2SO4, 3% PEG 400 in 0.1m NaKH2PO4 pH 7.5. The
PFE–inhibitor complex crystals were isomorphous with those for
[10] K. D. Green, M. Fridman, S. Garneau-Tsodikova, ChemBioChem 2009, 10,
2191–2194.
wild-type PFE (PDB ID: 1va4). The structure is high quality (Rwork
=
[11] U. T. Bornscheuer, R. J. Kazlauskas, Hydrolases in Organic Synthesis:
Regio- and Stereoselective Biotransformations, 2nd ed., Wiley-VCH, Wein-
heim, 2009, Chapter 5.
[12] C. C. Tseng, S. D. Bruner, R. M. Kohli, M. A. Marahiel, C. T. Walsh, S. A.
Sieber, Biochemistry 2002, 41, 13350–13359.
[13] K. K. Sharma, C. N. Boddy, Bioorg. Med. Chem. Lett. 2007, 17, 3034–
3037.
[14] 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. Verschue-
ren, Prot. Eng. 1992, 5, 197–211; M. Holmquist, Curr. Protein. Pept. Sci.
2000, 1, 209–235.
19.8%, Rfree =20.7%) with a resolution of 1.65 ꢁ, Table S3 in the
Supporting Information and PDB ID: 3ia2. The structure showed
only the R enantiomer of the sulfonyl, even though the inhibitor
was racemic. This R enantiomer corresponds the fast reacting
enantiomer of the corresponding ester substrate methyl (R)-2-
methylbutanoate.[47] PFE favors the R enantiomer of the ester 32-
fold over the S enantiomer and it is likely that PFE shows a similar
enantiopreference toward the nitrophenyl sulfonate enantiomers.
Molecular dynamics simulation: Starting from the D subunit of
the X-ray crystal structure of PFE containing the covalently bound
sulfonate (PDB ID: 3ia2), protons were added by using the protein
preparation wizard of Maestro (Schrçdinger LLC). By using the soft-
ware Desmond (D.E. Shaw Research, New York, NY), the protein
was placed in a box (69ꢂ69ꢂ69 ꢁ) of water molecules (SPC
model[48]) with 27 sodium cations and 20 chloride anions to neu-
tralize the charge. The system was initially minimized by using the
OLPS2005 force field[49] followed by several steps of equilibration
by using Berendsen thermostat and barstat method.[50] The simula-
tion (300 K, 1 atm pressure) by using Desmond for 1.2 ns with a
step time of 1 fs and recorded every 4.8 ps. The Coulomb interac-
tion was cutoff at 15 ꢁ and the long-range electrostatic interaction
was calculated with particle mesh Ewald (PME) method.
[15] U. G. Wagner, E. I. Petersen, H. Schwab, C. Kratky, Prot. Sci. 2002, 11,
467–478.
[16] X. Gao, X. Xie, I. Pashkov, M. R. Sawaya, J. Laidman, W. Zhang, R. Cacho,
T. O. Yeates, Y. Tang, Chem. Biol. 2009, 16, 1064–1074.
[17] L. Holm, J. Park, Bioinformatics 2000, 16, 566–567.
[18] G. D. Rose, L. M. Glerasch, J. A. Smith, Adv. Prot. Chem. 1985, 37, 1–109;
T. E. Creighton, Proteins: Structure and Molecular Properties, 2nd ed.,
pp. 225–227, Freeman, New York, 1993; C. M. Wilmot, J. M. Thornton, J.
Mol. Biol. 1988, 203, 221–232.
[19] J. Pleiss, M. Fisher, M. Peiker, C. Thiele, R. D. Schmid, J. Mol. Catal. B
2000, 10, 491–508; M. Fischer, J. Pleiss, Nucleic Acids Res. 2003, 31,
319–321.
[20] A. G. Murzin, S. E. Brenner, T. Hubbard, C. Chothia, J. Mol. Biol. 1995,
247, 536–540.
[21] B. Chakravarty, Z. Gu, S. S. Chirala, S. J. Wakil, F. A. Quiocho, Proc. Natl.
Acad. Sci. USA 2004, 101, 15567–15572.
[22] D. M. Lawson, U. Derewenda, L. Serre, S. Ferri, R. Szittner, Y. Wei, E. A.
Meighen, Z. S. Derewenda, Biochemistry 1994, 33, 9382–9388.
[23] K. K. Kim, H. K. Song, D. H. Shin, K. Y. Hwang, S. Choe, O. J. Yoo, S. W.
Suh, Structure 1997, 5, 1571–1584; K. H. Nam, S.-J. Kim, A. Priyadarshi,
H. S. Kim, K. Y. Hwang, Biochem. Biophys. Res. Commun. 2009, 389, 247–
250.
[24] S. A. Samel, B. Wagner, M. A. Marahiel, L. O. Essen, J. Mol. Biol. 2006,
359, 876–889.
[25] A. Mezzetti, J. D. Schrag, C. S. Cheong, R. J. Kazlauskas, Chem. Biol. 2005,
12, 427–437.
Acknowledgements
We thank the University of Minnesota, the Institute for Renewable
Energy and the Environment (University of Minnesota) and Natu-
ral Sciences and Engineering Research Council (Canada) for fi-
nancial support. We thank the Minnesota Supercomputing Insti-
tute for access to computers and molecular modeling software
and Dr. Nicholas Labello for help with the molecular dynamics
simulation.
[26] M. Nardini, D. A. Lang, K. Liebeton, K. E. Jaeger, B. W. Dijkstra, J. Biol.
Chem. 2000, 275, 31219–31225.
ChemBioChem 2011, 12, 768 – 776
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
775

R. J. Kazlauskas et al.
[27] M. J. Drçge, Y. L. Boersma, G. van Pouderoyen, T. E. Vrenken, C. J. Rueg-
geberg, M. T. Reetz, B. W. Dijkstra, W. J. Quax, ChemBioChem 2006, 7,
149–157.
[28] A. Roussel, N. Miled, L. Berti-Dupuis, M. Riviere, S. Spinelli, P. Berna, V.
Gruber, R. Verger, C. Cambillau, J. Biol. Chem. 2002, 277, 2266–2274.
[29] U. Derewenda, A. M. Brzozowski, D. M. Lawson, Z. S. Derewenda, Bio-
chemistry 1992, 31, 1532–1541.
[30] D. R. Ronning, T. Klabunde, G. S. Besra, V. D. Vissa, J. T. Belisle, J. C. Sac-
chettini, Nat. Struct. Biol. 2000, 7, 141–146.
[31] B. Pillai, M. M. Cherney, K. Hiraga, K. Takada, K. Oda, M. N. G. James, J.
Mol. Biol. 2007, 365, 343–361.
[40] M. W. Larsen, D. F. Zielinska, M. Martinelle, A. Hidalgo, L. J. Jensen, U. T.
Bornscheuer, K. Hult, ChemBioChem. 2010, 11, 796–801.
[41] E. R. Guinto, S. Caccia, T. Rose, K. Fꢃtterer, G. Waksman, E. Di Cera, Proc.
Natl. Acad. Sci. USA 1999, 96, 1852–1857.
[42] F. Stehle, W. Brandt, C. Milkowski, D. Strack, FEBS Lett. 2006, 580, 6366–
6374.
[43] Review: C. Milkowski, D. Strack, Phytochemistry 2004, 65, 517–524.
[44] P. Bernhardt, K. Hult, R. J. Kazlauskas, Angew. Chem. Intl. Ed. 2005, 44,
2742–2746; Angew. Chem. 2005, 117, 2802–2806; D. T. Yin, P. Bern-
hardt, K. L. Morley, Y. Jiang, J. D. Cheeseman, J. D. Schrag, R. J. Kazlaus-
kas, Biochemistry 2010, 49, 1931–1942.
[32] T. Collins, D. De Vos, A. Hoyoux, S. N. Savvides, C. Gerday, J. van Beeu-
men, G. Feller, J. Mol. Biol. 2005, 354, 425–435.
[33] A. Thomaeus, J. Carlsson, J. ꢁqvist, M. Widersten, Biochemistry 2007, 46,
2466–2479.
[34] A. Negri, E. Marco, J. Damborsky, F. Gago, J. Mol. Graphics Modell. 2007,
26, 643–651.
[35] G. Nicola, S. Peddi, M. E. Stefanova, R. A. Nicholas, W. G. Gutheil, C.
Davies, Biochemistry 2005, 44, 8207–8217.
[45] E. J. Corey, K. A. Cimprich, Tetrahedron Lett. 1992, 33, 4099–4102.
[46] J. D. Cheeseman, A. Tocilj, S. Park, J. D. Schrag, R. J. Kazlauskas, Acta
Crystallogr. D Biol. Crystallogr. 2004, 60, 1237–1243.
[47] S. Park, K. L. Morley, G. P. Horsman, M. Holmquist, K. Hult, R. J. Kazlaus-
kas, Chem. Biol. 2005, 12, 45–52.
[48] H. J. C. Berendsen, J. P. M. Postma, W. F. van Gunsteren, J. Hermans, In-
termolecular Forces (Ed.: B. Pullman), Reidel, Dordrecht, 1981, pp. 331–
342.
[36] R. Kourist, S. Bartsch, L. Fransson, K. Hult, U. T. Bornscheuer, ChemBio-
Chem 2008, 9, 67–69.
[49] G. A. Kaminski, R. A. Friesner, J. Tirado-Rives, W. L. Jorgensen, J. Phys.
Chem. B 2001, 105, 6474–6487.
[37] Z. Li, T. Lazaridis, Phys. Chem. Chem. Phys. 2007, 9, 573–581.
[38] P. Grochulski, F. Bouthillier, R. J. Kazlauskas, A. N. Serreqi, J. D. Schrag, E.
Ziomek, M. Cygler, Biochemistry 1994, 33, 3494–3500.
[39] R. C. Wilmouth, I. J. Clifton, C. V. Robinson, P. L. Roach, R. T. Aplin, N. J.
Westwood, J. Hajdu, C. J. Schofield, Nat. Struct. Biol. 1997, 4, 456–462;
P. A. Wright, R. C. Wilmouth, I. J. Clifton, C. J. Schofield, Eur. J. Biochem.
2001, 268, 2969–2974; E. S. Radisky, J. M. Lee, C.-J. K. Lu, D. E. Kosh-
land, Jr., Proc. Natl. Acad. Sci. USA 2006, 103, 6835–6840.
[50] H. J. C. Berendsen, J. P. M. Postma, W. F. van Gunsteren, A. Dinola, J. R.
Haak, J. Chem. Phys. 1984, 81, 3684–3690.
Received: November 15, 2010
Published online on February 23, 2011
776
www.chembiochem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2011, 12, 768 – 776